Dear SingularityNET Community,
During 2024, SingularityNET made huge strides forward towards Artificial General Intelligence (AGI) and achieving our mission, laying the foundation to lead in the AGI era. We continued to transform our Foundation, innovate, move forward on our AGI R&D roadmap, build the decentralized infrastructure for the global economy, embrace open collaboration, and guide the development of neural-symbolic AGI in a beneficial direction.
We formed the Artificial Superintelligence (ASI) Alliance, the largest open-source, independent AI R&D initiative, free from corporate and governmental control. Through this collaboration between SingularityNET, Fetch.ai, Ocean Protocol and CUDOS, the Alliance provides an open, decentralized technology stack, for the research, development, and commercialization of artificial Intelligence (AI), offering a framework that will power the future of AGI and beyond.
We advanced the fundamental science of several critical technologies. We announced the Alpha release of the open-source OpenCog Hyperon AGI framework, scaled the interpretation and compilation of the MeTTa language for cognitive computations, introduced the MettaCycle Layer 0++ technology for secure, scalable, and performant decentralized computing, and accelerated the development of AGI-optimized software and hardware architectures.
R&D also introduced AIRIS (Autonomous Intelligent Reinforcement Inferred Symbolism), one of the first intelligent systems capable of autonomous, adaptive learning with real-world applicability. We deployed AIRIS in the Minecraft 3D gaming environment to help increase the generality of autonomous agents and the data efficiency of learning goal-directed behaviors.
Our Platform development efforts advanced the foundation for decentralized AI infrastructure, establishing comprehensive frameworks for next-generation AI services. Our teams delivered significant enhancements across the ecosystem, from sophisticated marketplace capabilities and training frameworks to advanced developer tools and streamlined onboarding systems. These developments, combined with multi-chain integration and enhanced security protocols, reinforce our vision of an accessible, decentralized AI ecosystem. This ecosystem will serve as the decentralized infrastructure for the global economy as the world enters the AGI era.
We invested in energy-efficient and sustainable hardware infrastructure to accelerate our AGI and ASI development. In June, we committed $53 million to establish the first modular supercomputer dedicated to decentralized AGI evolution, alongside state-of-the-art High-Performance Computing facilities. Our partnership with Tenstorrent further strengthens this foundation, enabling the development of specialized AGI-optimized chip architectures to advance our hardware capabilities.
Our ecosystem projects demonstrated their abilities to transform research into vertical and domain-specific solutions across decentralized finance, robotics, biotech and longevity, gaming and media, arts and entertainment, and enterprise-level AI. These advances are poised to multiply the impact of our AI technologies across society and businesses.
Recognizing that AI and AGI stewardship should not be confined within Big Tech or government walls, we conducted ethical reflection, launched programs and offered grants to aid and accelerate the growth of open-source decentralized initiatives, and fostered responsible AI practices across our ecosystem. In February 2024, we held the inaugural Beneficial AGI Summit in Panama to discuss how to guide AGI governance and leverage generally intelligent machines to support ethical, social, and economic progress.
Our perspective on AGI is clear and consistent: AGI needs to be in the hands of humanity at large — to be controlled by no one person or organization and enjoyed by everyone. That is why we will continue to provide practical solutions to help accelerate progress toward decentralized beneficial AGI. We will actively expand our involvement with wider stakeholders to mitigate the socio-ethical risks of frontier technologies, bridge the divides between the developed and developing world, and reconcile technological, economic, and social progress with environmental sustainability.
We enter 2025 more committed to building beneficial AGI and ASI than ever, more strategically focused, technologically capable, and positioned for massive acceleration and scaling.
Janet Adams, COO of SingularityNET
In 2024, we progressed the technical development of the Hyperon framework for AGI and beyond, transitioning from initial proof-of-concept and prototyping phases to establishing the framework as a functional software platform with defined product development trajectories.
In April, the Alpha release of the MeTTa cognitive language, a core component of Hyperon, marked a significant milestone in Hyperon’s development. This release coincided with the formation of an open-source community of Hyperon developers and MeTTa programmers.
To support community development, ongoing biweekly collaborative sessions (MeTTa Coders) continue to address two primary domains: technical discussions on MeTTa programming and strategic meetings focused on AGI development using the Hyperon framework. Through the Deep Funding (DF) initiative, we offered $1.25m in grants to foster this community, grow the adoption of MeTTa, and advance beneficial AGI R&D via the Hyperon framework.
At the 17th AGI Conference (AGI-24) in Seattle, a Hyperon Workshop offered participants hands-on experience with the MeTTa language, covering both its fundamentals and advanced application. The workshop also showcased a range of Hyperon applications, with presentations on PRIMUS, Neoterics, Mind Children, PLN/Pattern Matcher, Rejuve.Bio’s Neural-symbolic Hypothesis Generation, MeTTa-NARS, NACE, and more.
Following the Alpha release, the Hyperon framework advanced through several key development phases during 2024, pursuing AGI through an open, decentralized approach that leverages blockchain and distributed networks to foster collaboration and ensure equitable access to frontier technologies.
The Alpha release of the MeTTa interpreter prioritized usability, incorporating comprehensive documentation and enhanced accessibility features. The existing code documentation was expanded, and functionality for managing documentation directly within MeTTa was integrated. Tutorials were published on the metta-lang.dev web portal, along with an online interpreter that provides executable examples and a sandbox environment, enabling users to explore the language without requiring local installation.
MeTTa was published as a Hyperon package on PyPI, offering support to multiple platforms, including Apple silicon, and thereby simplifying installation for Python users. The implementation of the interpreter in Minimal MeTTa (a concise set of basic instructions used to describe the interpreter) was fully aligned with the initial ad hoc Rust implementation, leading to more simplified and stable semantics. The core subset of the standard library was refined to minimize potential backward compatibility issues, and the Alpha release included a module system to allow out-of-the-box usage of additional MeTTa libraries.
Following this release, development continued on these and other components, including the expansion of the stdlib functionality, publication of new tutorials, and a significant merger of the Rust-MeTTa and Minimal MeTTa interpreters. This merger yielded a 50x performance improvement for the Minimal MeTTa version. Another notable technical achievement was the implementation of universal access to Python modules and classes, allowing direct method calls from MeTTa without the need for additional wrappers.
These comprehensive enhancements significantly matured the interpreter, driving its advancement in conjunction with the development of other core Hyperon components.
The Integration of the Distributed Atomspace (DAS) with MeTTa was initiated around the time of the Alpha release. This integration necessitated the development of missing DAS functionality and a dedicated DAS adapter in MeTTa. This adapter, responsible for converting MeTTa code into DAS queries and interpreting the results back in MeTTa, marked the first module developed in MeTTa beyond the standard library.
Subsequent DAS development focused on implementing an innovative cache subsystem that integrates with the Query Engine API and AttentionBroker to set Short and Long Term Importance. This approach diverges from traditional caching by prioritizing the relevance of query results through sorting and partitioning, which offers significant advantages for AI agents conducting combinatorial searches. The implementation leveraged a multi-database approach, routing queries across backends like MongoDB and Redis to optimize performance and scalability. The successful deployment of the DAS on NuNet nodes by year end demonstrated its practical functionality within distributed computing environments.
The OpenCog Hyperon framework for AGI R&D supports multiple cognitive architectures including one-off question-answering chat systems, theorem provers, and even cognitive architectures designed to emulate aspects of the human mind. However, a particular cognitive architecture, namely CogPrime has remained central to the development of OpenCog Classic.
In 2024, this trajectory shifted with the design of PRIMUS, a next-generation architecture intended for implementation within Hyperon. PRIMUS builds upon the foundation of CogPrime, incorporating both the unique capabilities of Hyperon and the latest advancements in AGI and neural-symbolic AI.
PRIMUS will continue to incorporate subsystems for uncertain inference, including Probabilistic Logic Networks (PLN) and Non-Axiomatic Reasoning System (NARS), evolutionary methods such as MOSES, attention and resource allocation, motivation, goal generation, perception, action, concept formation, support tools, and more. Driven by curiosity, PRIMUS will actively experiment with and explore its environment, leveraging causal and temporal reasoning to construct and refine its internal world model.
While incorporating recent AI advancements (e.g., LLMs) to enhance its internal modules, the dynamics of PRIMUS will remain guided by the principle of cognitive synergy. This principle dictates that the cognitive processes within PRIMUS will collaborate rather than conflict. If one process encounters an obstacle, it can translate its intermediate state into a format understandable by other processes and solicit their assistance. This will result in an emergent intelligence that surpasses the capabilities of its individual parts, fostering continuous evolution.
PLN is a logic designed to emulate human-like common sense reasoning. This logic excels at handling uncertainty, empirical and abstract knowledge, and the relationships between them. While closely resembling Non-Axiomatic Logic (NAL), PLN is firmly grounded in probability theory. For example, the notion of uncertainty is captured using second-order probability distributions. This logic provides a common language that contributes to the cognitive synergy.
As an early test case for MeTTa implementation, PLN became the first cognitive component targeted for integration long before the Alpha release. This resulted in the exploration of MeTTa’s idiomatic constructions and revealed missing features. The porting process was approached from different angles: Proofs as match queries, Proofs as custom Atom structures, Proofs as programs, and properties as dependent types. Currently, the latter represents the most advanced approach.
Central to the dependent type approach is the implementation of a generic program synthesizer. Program synthesis experiments with uncurried backward chaining were handled. Furthermore, to implement a constructive version of the Pattern Miner, function calls within logical rules were introduced as a chainer extension.
Different versions of the forwards and backward chainers in MeTTa were prototyped and experimented with both using proofs as programs and proofs as program execution traces:
Inference control experiments including Curried Backward Chainer, Controlled Backward Chainer, Maximum Depth Controlled Backward Chainer, Reasoning-base Controlled Backward Chainer, and subtyping inference experiments were conducted.
A notable PLN application is mitigating LLM hallucinations. This is achieved by asking an LLM to translate natural language statements into the logical (PLN) representation, either directly or using Lojban as a proxy. PLN can verify the internal consistency of statements and fact-check them against a KG. A functional prototype of this system, demonstrating improvements in LLM question-answering through hallucination reduction, was presented during our technical Hyperon meetup in Brazil.
The ECAN implementation in MeTTa has been progressing through two parallel tasks: porting the previous C++ attention bank implementation into MeTTa and fixing a previous experimental implementation in C++.
MOSES is an evolutionary algorithm designed to evolve computer programs by optimizing them toward a specific goal or fitness function. MOSES uses a probabilistic modeling technique called Estimation of Distribution Algorithms (EDA) to build and refine populations of programs. It searches for optimal or near-optimal solutions by exploring the space of program structures in a more guided and efficient way, leveraging probabilistic models to predict and generate promising new program candidates. This makes it suitable for complex, symbolic tasks in the pursuit of AGI.
MOSES port from OpenCog Classic to MeTTa has started. SG-GP (Stochastic Grammar Genetic Programming) and SGE (Structured Grammar Evolution) implementations are in progress.
At the beginning of 2024, the port of NARS to MeTTa was enabled by the MeTTa-Morph extension, which compiles some computations in MeTTa using a Scheme compiler. This technical advancement facilitated NARS demonstration across multiple application domains.
The NARS+GPT project was ported from OpenNARS to MeTTa with the use of the MeTTa-Motto library for communicating with LLMs. Testing MeTTa-NARS+LLMs on a question-answering benchmark yielded promising results.
The Non Axiomatic Casual Explorer (NACE) navigational system was implemented with the use of MeTTa. “Explorer 1” design for NACE was introduced, resulting in the Integrated Hybrid System (NARCES) featuring semantic to sub-symbolic interconnection in NARS.
The experiment with Non-Axiomatic Logic truth values for Signal Temporal Logic formulas laid the groundwork for future research with Integrated Hybrid System to feature semantic to sub-symbolic interconnect.
The NARCES framework was integrated with the Robotic Operation System (ROS2), resulting in a live demonstration of robotic manipulator control within a simulated environment. This demonstration highlighted the framework’s ability to integrate complex cognitive processes.
The evolution of the MeTTa interpreter has enabled the development of key libraries, including the SingularityNET Platform software development kit (SDK) in MeTTa and MeTTa-Motto, laying the groundwork for a Hyperon-powered Knowledge Layer.
In 2024, the SingularityNET Platform SDK was released as a standard MeTTa module, distributed alongside the MeTTa interpreter. This provides comprehensive Platform functionality accessible through MeTTa programs, including listing organizations and services from the Platform registry, executing free calls, managing payment channels, and processing deposits for paid calls.
By automatically generating MeTTa functions for service calls and abstracting protobuf details, the MeTTa module streamlines Platform service calls to be as straightforward as invoking standard MeTTa functions. This capability enables service specification design in MeTTa and provides a pathway for utilizing these services as Hyperon plugins.
Concurrently, AI-DSL demonstrated significant progress in automatically identifying suitable services and composing them based on provided task specifications.
Our MeTTa-Motto library, initiated in late 2023 as a wrapper for calling ChatGPT from MeTTa, facilitates the integration of LLMs with MeTTa, allowing developers to leverage the capabilities of both technologies. This integration enables the development of AI applications that utilize natural language processing (NLP) and symbolic reasoning.
MeTTa-Motto recently expanded to include integrations with OpenRouter with proxy access to multiple LLMs and Anthropic API, LangChain infrastructure, vector embeddings for retrieval-augmented generation, and ontological systems like Wikidata and DBpedia.
The key technical distinction of MeTTa-Motto lies in its ability to bridge LLM interactions with MeTTa’s logical processing. This enables several capabilities: composing and chaining LLM calls using MeTTa scripts; processing LLM outputs as MeTTa expressions for further reasoning; using MeTTa’s symbolic knowledge to construct targeted prompts; managing stateful dialogues through specialized agents; and accessing external tools and data sources through LangChain integration.
A notable achievement was the implementation of native adapters for calling LLMs and KGs Platform services in MeTTa-Motto, using the SingularityNET SDK in MeTTa. This development paves the way for the Knowledge Layer on the Platform with access to KGs and LLMs from Hyperon within a unified framework.
Another significant trend in MeTTa-Motto’s evolution is the incubation of an agent framework. This framework is emerging as a standalone MeTTa library with multi-threaded event processing capabilities. Its design supports not only conversational agents with features like interruption handling and speech-canceling but also extends to robotic control.
The introduction of Hyperon as a platform for AGI R&D and practical applications highlighted the importance of ongoing efforts to enhance performance and scalability. Building upon the foundational work undertaken in 2024 across multiple complementary approaches, addressing the fundamental challenges of AGI scalability remains a primary focus for 2025.
These parallel development tracks embody the project’s open and decentralized nature and reflect the complex requirements of a language designed to serve as an integration point between diverse algorithmic paradigms, including evolutionary algorithms, symbolic logical inference, and neural systems.
In 2024, we initiated the development of a MeTTa-to-JVM (Jetta) compiler in Kotlin. The compiler is initially designed to cover a subset of the MeTTa language, including custom functions operating on grounded types such as numbers, strings, booleans, and others. Currently, the compiler supports the compilation of simply typed functional programs.
A compiler server and an adapter were implemented within the MeTTa interpreter via its application programming interface (API), enabling MeTTa programs to send specific code segments for compilation and subsequently utilize the compiled functions during interpretation. This approach has demonstrated performance improvements approaching a million-fold speedup for functions such as factorial calculations.
The compiler has been further enhanced to support lambda expressions and higher-order functions. This capability allows MeTTa programs to compile code generated at runtime, a crucial feature for speeding up genetic programming and other program synthesis methods.
The recent introduction of nondeterminism parameterized with map functions to Jetta represents a significant advancement, enabling more flexible and powerful computational capabilities. This is a core aspect of the MeTTa language given the nondeterministic nature of pattern matching.
With these advancements in performance optimization and its potential to function as a pluggable component within the MeTTa ecosystem, Jetta demonstrates considerable promise for future development and broader applications.
A parallel initiative, the MeTTa Optimal Reduction Kernel (MORK), focuses on an algebraic approach to composing Atomspace-wide operations. This implementation is based on a specialized, zipper-based, multi-threaded virtual machine, which provides native and highly scalable parallelism. By rearchitecting certain Hyperon bottlenecks, MORK has the potential to achieve performance improvements ranging from thousands to millions of times for key use cases.
Current results include a draft PLN implementation built on top of MORK and a demonstration of loading 1 billion MeTTa atoms representing a biomedical dataset. MORK also features a batched search for the closest hypervectors representing MeTTa atoms, which is currently a bottleneck under investigation.
Another parallel development track is MeTTaLog, which implements a MeTTa transpiler built on the Warren Abstract Machine in Prolog. This approach leverages Prolog’s efficient unification and inference engine. MeTTaLog boasts the highest compatibility with the MeTTa interpreter among the parallel efforts. This includes a functional PLN implementation within MeTTaLog and partially complete integration with MeTTa libraries. Furthermore, MeTTaLog demonstrates the capacity to load and query a billion atoms, exhibiting speedup in certain cases.
Recent demonstrations of MeTTaLog have shown promising performance improvements, particularly in concurrent execution and data structure optimization. These parallel development tracks, including Jetta, MORK, and MeTTaLog, continue to progress, each offering unique solutions to the challenges inherent in building a scalable AGI “language of thought.”
We deployed the AIRIS experiential learning system within the Minecraft 3D gaming environment, marking one of the first intelligent systems capable of autonomous, adaptive learning. This development will leverage Fetch.ai’s agent technologies, CUDOS’ compute infrastructure integration, and Ocean Protocol’s data frameworks for memory management, creating comprehensive individualized user experiences.
The system implementation demonstrates significant capabilities in dynamic navigation, obstacle adaptation, and efficient pathfinding protocols. AIRIS architecture enables real-time environmental adaptation without extensive retraining requirements, implementing dynamic rule generation through partial observation frameworks. The system’s functionality extends across complex terrain management including water bodies and cave systems, while maintaining computational efficiency for real-time rule application.
The development roadmap focuses on enhanced object interaction capabilities, multi-agent collaboration frameworks, and advanced reasoning protocols. System architecture aims to expand beyond navigation to include sophisticated decision-making frameworks for contextual tasks, social dynamics, and strategic problem-solving. These implementations establish foundations for real-world applications across autonomous robotics, smart assistance systems, and adaptive learning frameworks.
The AIRIS architecture consists of four primary components: the Rule Base for learned rules, the State Graph containing the world model of current and future predicted states, and Pre-Action and Post-Action Observation sequences. The system builds sets of causal rules from observations of environmental changes, enabling autonomous learning without pre-specified logical knowledge or encodings.
The system’s State Graph creation implements Focus Values compilation from state changes, reducing computational complexity through targeted attention mechanisms. The system evaluates rules through precondition match scoring (0:1) and priority queue management, with State Confidence metrics for prediction validation. This architecture enables generalization capabilities through partial state representation rather than complete state transitions.
Key achievements in testing environments demonstrate AIRIS’s efficient learning capabilities, with the system requiring significantly fewer training iterations than traditional reinforcement learning approaches. The system’s ability to adapt objectives without efficacy impact and transparent rule generation provides advantages for autonomous agent development. These capabilities position AIRIS as an innovative approach to causality-based AI, moving beyond traditional expert systems while maintaining scrutability and operational efficiency.
In 2024, we launched a $1.25M R&D grant program through our Deep Funding initiative to advance decentralized, benevolent and beneficial AGI. The program comprised 14 Requests For Proposals (RFPs) focused on advancing the OpenCog Hyperon framework for AGI at the human level and beyond, representing a strategic approach to open decentralized AI R&D.
The RFPs generated considerable global interest, with 124 proposals submitted from individuals and teams worldwide, highlighting the growing AI community engagement with our hybrid neural-symbolic approach to AGI R&D and vision of open decentralized AGI. The proposals covered a diverse range of research areas, including the use of LLMs for the Meta-Optimizing Semantic Evolutionary Search (MOSES) Evolutionary Algorithm, neural-symbolic deep neural network (DNN) architectures, clustering heuristics in the MeTTa programming language, AGI motivation systems, probabilistic logic network guidance, quantum computing, and AGI-specific hardware configurations.
The RFP evaluation process assessed each proposal’s technical strength, potential for integration, and contribution to our broader AGI strategy. Experts were assigned to RFPs based on their specific areas of expertise. After individual assessments were completed, all reviews were compiled and shared among the experts. Final reviews were conducted by teams of senior technologists, each composed of members with RFP-specific expertise. To make the process completely transparent, we have made all proposals and individual reviews publicly available.
Our grant program prioritized open-source development principles, requiring comprehensive documentation, public code repositories, and detailed technical reports. This approach created a robust collaborative platform, enabling researchers globally to advance cutting-edge AI technologies and reinforcing our commitment to open, decentralized, collaborative AGI R&D.
For details on the awarded projects, please see our official announcement blog.
During 2024, we advanced our decentralized AI Platform development, further augmenting our comprehensive infrastructure for AI service creation and deployment, focusing on decentralization, safety, scalability, and accessibility.
Early in the year, the Platform underwent foundational architectural refinements through environment synchronization between production and testing systems, expanded HTTP type services, and the integration of AI model training capability.
These developments established a robust decentralized AI infrastructure that aggregates multiple AI services on a single decentralized network, and functions as an end-to-end ecosystem of diverse, complex processes and interconnected components. Enhanced user interface architecture, optimized onboarding protocols, and comprehensive documentation frameworks support this infrastructure.
These core functionalities serve as the foundation, supplemented by ongoing R&D initiatives exploring the implementation of next-generation tools and third-party integrations.
The Developer Portal underwent significant technical advancement, transitioning to TypeScript and implementing dual-mode search architecture. The documentation was expanded to include comprehensive SDK integration guides, API specifications, and technical references.
The Portal advanced through improved navigation with section-specific routing, optimized resource loading, and enhanced mobile accessibility. The system integrated Vitepress on TypeScript, enabling new layout components in document content, structural refinements, and accelerated Portal development capabilities. Search functionality implementation enables user-selectable display modes, providing full text line presentation of search results across both document titles and content.
The Portal’s design incorporates comprehensive UI updates including theme-changing functionality with dual content interaction modes: single-branch mode for collapsing open subsections during navigation and all-branches mode for complete category subsection visibility. Additional features include collapsible content menus, user feedback form implementation for bug reporting and feature requests, expandable image viewing, and revised documentation.
Our decentralized AI Marketplace underwent continuous enhancement, including implementing a Data Preset for service fine-tuning. The Platform integrated advanced dataset analysis capabilities, visualization tools, and improved dataset management features.
Payment architecture improved through enhanced Metamask integration, with notification service implementation including field validation. System updates incorporated newer library versions, while Metamask connections now automatically request network or account changes. Payment systems underwent refinement to reduce the number of required signatures.
Training functionality received a complete overhaul, enabling user-driven model creation and service training on custom datasets. The system expanded with comprehensive dataset processing capabilities, including analysis, merging, cleaning, and optimization functions. Newly introduced updates included fixes for authorization, element and route display in the application, and updated validation for authorization forms.
The SDK ecosystem advanced through multi-language development. JavaScript SDK architecture underwent restructuring into Node.js, Web JS, and Common Core components, with new module implementations for payment strategy protocols, model management, and service interaction. System updates resolved transaction formation issues affecting Publisher and Marketplace operations.
Architecture refinements improved client service creation with predefined payment strategies, enabling selective free and paid AI service calls. System optimization included free call functionality fixes and codebase enhancements supporting training operations. Development focuses on writing tests, logging system updates, and Filecoin integration protocols.
The Platform expanded through Boilerplate application deployment, providing Marketplace-independent service access capabilities. This implementation delivers integrated UI frameworks for service calls, wallet connectivity, and FET (ASI) acquisition. Onramp integration in Boilerplate enhanced payment option accessibility for service utilization. Comprehensive documentation provides implementation guidance for Boilerplate deployment and custom UI development.
The Python SDK advanced through the integration of Training v2 protocols, network parameter configuration systems, expanded proto-compilation options, and enhanced test coverage frameworks. Development included automated PROTO file generation for Marketplace service calls and SDK-CLI dependency separation. Proto file management enables current service proto generation through force mode implementations.
Platform capabilities expanded to support Python 3.10 and 3.11, incorporating comprehensive refactoring and dependency updates. System enhancements delivered custom organization and service listing methods, simplified config setup protocols, and automated proto-file parsing for service method detection. Infrastructure improvements included detailed internal documentation frameworks and channel caching implementation for optimized paid service access. Development addressed security dependency protocols and introduced Filecoin functionality, with multiple releases deploying system improvements and refinements.
The Daemon component advanced from version 5.1.2 to 5.1.5, implementing Node.js dependency removal, dynamic SSL certificate update protocols, and Zap logging system integration. Logging performance was enhanced by a factor of 1,000 while reducing memory allocation by 10x. Additionally, embedded etcd logging received configurable management capabilities.
Blockchain functionality expanded through new contract listener implementation for organization metadata change detection and dynamic etcd endpoint reconnection. System updates included deprecated method removal and library updates for EVM and IPFS interactions. Feature implementation delivered configurable free call parameters and simplified configuration protocols for accelerated onboarding.
The Platform integrated custom error codes, enhanced training capabilities, and comprehensive Filecoin and Lighthouse support. Infrastructure development included unit test refactoring and new integration testing for gRPC request proxy methods between client and provider services. CI/CD framework underwent optimization for enhanced automation protocols.
Training V2 development delivered refined training methodologies and new architectural implementations. System updates incorporated a new “proto” file design with enhanced AI model communication protocols, with multiple releases deploying these capabilities throughout the development cycle.
The Publisher Portal advanced through automated UI protocols, enhanced daemon interaction frameworks, and comprehensive onboarding system integration. Platform accessibility expanded through new configuration tools and detailed documentation implementations.
Service and organization registration functionality restored core pipeline operations and introduced multi-organization management capabilities. System optimization included organization duplication verification and streamlined service registration pathways with integrated navigation protocols.
Infrastructure updates delivered enhanced organization service displays featuring pricing and daemon installation specifications. Platform expansion included rapid configuration generation through simplified field entry systems, integrated FAQ frameworks, and multi-format onboarding guidance. System refinements incorporated field validation protocols and optimized marketplace service assembly for accelerated deployment.
The Text User Interface (TUI) advanced through comprehensive design refinements, expanded functionality, and enhanced documentation. Notable developments included Python installation automation protocols, advanced navigation capabilities, and Filecoin integration. The Platform incorporates detailed parameter documentation and streamlined operational guidance for common scenarios.
The Platform Development team supported the Cardano Chang Hardfork integration, implementing comprehensive updates across Cardano node architectures, dbsync implementations, Cardano service protocols, bridge script frameworks, token transfer systems, and staking mechanisms to maintain blockchain compatibility standards.
The UI Sandbox represents a significant advancement in AI service interface development within the SingularityNET ecosystem. This sophisticated development environment addresses historical challenges in AI UI creation, enabling researchers and developers to focus on core algorithm development while ensuring effective user interaction capabilities.
The system architecture delivers comprehensive development tools including file management protocols, compilation systems, preview capabilities, environment variable management, and error-handling mechanisms. Core features include dedicated code development and result visualization areas, incorporating demo mode functionality, frontend compilation modules, and extensively documented element libraries. The Custom Components Library and enhanced project management tools streamline service development processes across the platform.
The UI Sandbox includes feedback systems, FAQ frameworks, and publisher-compatible frontend deployment with existing frontend modification support. The Sandbox implements rapid UI prototyping and deployment capabilities, reducing development time while accelerating AI service adoption. Current development focuses on template library architecture for zero-code frontend assembly, responding to community and developer feedback through systematic improvement cycles.
The Command Line Interface (CLI) underwent systematic enhancement through legacy argument removal, private key and mnemonic encryption protocols integration, and documentation expansion. System functionality advanced through channel caching optimization and comprehensive Filecoin Storage Provider integration.
Filecoin integration established complete file management capabilities through Lighthouse implementation, while enhanced system performance delivered through channel caching protocols reduced command execution times. Security frameworks expanded through private key and mnemonic phrase encryption, complemented by metadata and value validation systems. Other developments included automated documentation generation updates and functional testing implementation covering 30% of command protocols, with multiple releases deploying these fixes and capabilities.
As part of the DeepFunding initiative, we are developing technical requirements and designing the architecture for a comprehensive system centered around an escrow smart contract. This contract will facilitate milestone-based payments to project teams, ensuring secure and transparent fund management.
Key features include partner escrows, milestone-based payments with multi-signature approval, and flexible payment options in stablecoins or tokens. This system is designed to enhance operational efficiency, promote financial security, and foster trust. We are also updating the Platform’s MPE contract to add new functions, such as adding multiple users to one channel and working on subscription model mechanics.
The revamped Voting Portal introduces enhanced functionality through a tabbed interface, enabling voters to monitor realtime voting progress and verify their voting power. Transparency improvements include visible vote signatures and robust validation mechanisms, with downloadable voting results and balance snapshot data. The system offers comprehensive wallet integration spanning Ethereum, Binance, and Cardano networks, supporting software and hardware wallet connections.
The Token Bridge introduces a modernized interface incorporating GameFi design elements, providing users with positive step completion feedback. It integrates TradingView rate data visualization, conversion history filtering capabilities, and enhanced wallet connection architecture. User experience improvements include guided navigation elements and upfront fee calculations.
The Bridge infrastructure expanded with support for ASI (FET) and WMTx tokens, enabling bi-directional transfers between Ethereum and Cardano networks. Implemented functionality includes ASI cross-chain transfers, native AGIX to ASI migration on Cardano, and WMTx inter-chain conversions. The system incorporates comprehensive token management features including locked tokens conversion support, liquidity pool transfers, fee collection protocols, and conversion ratio frameworks.
As part of the first phase of migrating payment channels to the Cardano blockchain, we prepared a Request for Proposal (RFP) for adapting the MultiParty Escrow (MPE) smart contract. The goal is to move the existing Ethereum-based MPE functionality to Cardano, using its Extended UTxO model. This will enable secure and decentralized transactions between SingularityNET clients and AI service providers. The RFP details how the payment channels will work, including features like token deposits, channel claims, expiration management, and extensions. The implementation will follow Cardano’s UTxO architecture while ensuring the system is scalable, secure, and fits into SingularityNET’s marketplace.
Concurrently, we have been enhancing Cardano support on the Platform, notably SingularityNET staking. Due to the recent Chang hard fork, this required updates to the Cardano Transaction Library (CTL). Updates remain in development to align with new system specifications with upcoming milestones including the finalization of the MPE smart contract and the completion of the new staking system.
For the migration from Cardano Native Token (CNT) AGIX to CNT FET (ASI), essential bridge mechanics for tokens with shared ownership were implemented. This includes a shared ownership token contract and liquidity contract for custodial token transfers, a secure multi-signature approach with external transaction signer services, and the ability to create conversions with a given conversion ratio.
These developments, along with continuous refinements in dataset preprocessing, training model management, and blockchain integration, have significantly advanced our Platform’s capabilities, establishing a robust foundation for future innovations.
The SingularityNET AI Music Lab has advanced significantly in developing an AI-powered music generation system, completing initial development phases to establish foundations for innovative music creation tools.
The initiative began with a comprehensive data acquisition phase, accumulating 170,000 hours of audio content to serve as the foundational training dataset for AI model development. The data infrastructure expanded through systematic metadata extraction, implementing mood and genre classification systems, live audio identification protocols, and NLP-based metadata analysis frameworks.
The development advanced through a comprehensive evaluation of music analysis tools, conducting systematic testing of vocal separation capabilities, structure analysis systems, and metadata assessment frameworks. The Platform implemented defined testing criteria and benchmarking protocols using validated sample datasets. Research initiatives focused on model parameter optimization to enhance generation quality and user control mechanisms.
Current development centers on pilot training and evaluation protocols, establishing foundations for small-scale generation models through baseline system implementation, audio preprocessing frameworks, and checkpoint validation using targeted data subsets.
The technical expansion will focus on medium-scale production scaling, including complete data processing implementation, genre-based classification systems, and robust user checkpoint validation. The model will be released with comprehensive documentation and deployment frameworks.
Parallel development continues in music stem separation technologies, advancing capabilities for vocal and instrumental isolation to enhance dataset quality and support downstream music analysis and generation tasks.
Our AI NLP Lab has made significant progress in LLM research, Generative Language Model (GLM) service development, and the creation of sophisticated dialogue systems.
The Lab’s LLM research direction has focused on fine-tuning and optimizing LLMs for various applications. The team has explored distributed fine-tuning techniques, including Tensor Parallel and Pipeline Parallel with LoRA, to enhance model training efficiency and performance. Additionally, they have investigated methods for utilizing low-resource hardware for large model training, making powerful AI capabilities more accessible.
As for GLM service development, the Lab has designed and deployed services on the SingularityNET AI Marketplace. The team has selected and optimized models and released two versions of the GLM service, each offering enhanced capabilities and performance.
Furthermore, the Lab has successfully trained an Arabic-English translation model, demonstrating its expertise in cross-lingual language processing. The team collected and prepared data, set up a robust cluster environment, and trained the model, achieving significant results.
The AI NLP Lab has also tested cutting-edge hardware, including Tenstorrent T3000. Through this rigorous testing and benchmarking, the team has evaluated the hardware’s capabilities for running LLMs, providing valuable insights for optimizing AI inference.
The Lab has also made strides in developing Fine-tuning as a Service (FTaaS), enabling users to customize language models for specific tasks and domains. The team has designed and implemented the core service architecture, developed a proof-of-concept LLM service, and created an instruction showcase to demonstrate the capabilities of FTaaS.
One of the Lab’s notable achievements is the development of the TWIN dialogue system, an AI-powered system capable of engaging in human-like conversations. The team has conducted extensive R&D, fine-tuning models on various datasets, incorporating advanced techniques such as graph traversal and style modeling, and optimizing the system for high-speed performance. The TWIN system has been successfully deployed in production and is being integrated with various applications.
The AI NLP Lab has also contributed to the development of the SingularityNET Assistant, an AI-powered assistant that provides information and support to users of the SingularityNET Platform. The team has explored various approaches to improve the assistant’s capabilities, including the MeTTa approach, and has developed and released a new version of the system with enhanced features.
The team has also upgraded its previous project in the Rejuve Automated Insights Generator, contributing to system improvement, prototype development, and the generation of new insights.
Finally, the AI NLP Lab has actively contributed to the advancement of the SophiaVerse project, designing the overall system architecture and developing various modules, including the FreeForm Input module, Dialog Core logic, and Semi-Freeform Output module. The team has also researched different embedding systems and developed an aligned output module for the FreeForm Input and a FreeForm Output module prototype development.
Deep Funding is a community-driven program that has been created to support our mission of developing decentralized, democratic, inclusive, beneficial AGI. The main goal of DF is to support the growth of our decentralized AI Platform.
The implementation of a custom-built DF portal in 2024 enabled transition from third-party platform dependency to a comprehensive RFP-based funding program. With 15 RFPs currently in progress, the platform demonstrates significant expansion of funding capabilities through its new infrastructure.
Technical development advanced through the launch of community.deepfunding.ai reflecting the growing maturity of the Circle-based organizational structure. Community-driven initiatives marked key achievements, including the Developer Outreach Circle’s first grassroots-developed hackathon deployment.
The January 2025 roadmap includes comprehensive platform updates: homepage redesign, RFP category page implementation, Community Hubs initiative launch for community growth and diversification, DFR5 alignment with ASI: Create platform development, and RFP-Create flow design presentation for community-driven ideation.
Key developments through 2025 will deliver RFP-Create flow implementation with ideation campaigns, Open Challenges funding initiative, expanded hackathon capabilities, and enhanced community website features. Technical improvements include Single Sign On (SSO) integration connecting the community site with the portal and dedicated Mattermost environment, alongside advanced portal features with reputation system integration through richer profiles. Backend infrastructure will advance through operations.deepfunding.ai automation processes, while partnership frameworks expand for joint RFPs and Pools.
Additional platform developments and features will emerge through community engagement and technical advancement. This systematic development approach establishes a robust infrastructure for continued growth in decentralized funding initiatives while maintaining a focus on community engagement and technical advancement.
Visit deepfunding.ai, the DF Telegram channel, or LinkedIn page to explore funding rounds, participation opportunities, and active projects shaping the SingularityNET decentralized AI Platform.
For the year, we laid the foundation for a decentralized AGI supercomputing ecosystem, focused on scalability, adaptability, and high-performance AI processing. Key initiatives included developing strategic infrastructure, forging of global partnerships, and optimizing AI workloads to support AGI applications at scale.
We launched a global strategy focused on the rapid expansion of a network of AGI supercomputing facilities. These facilities feature hardware configurations designed to surpass traditional GPU clusters, offering optimal efficiency for task-specific processing units, federated learning, and neural-symbolic AI running on Hyperon. By enabling continual learning, scalable inference, and hybrid AI architectures, these systems are laying the groundwork for AGI powered by next-generation compute infrastructure.
We adopted a multi-location compute strategy to ensure global scalability and redundancy. This network will leverage ecosystem partners for decentralized AI resource sharing and blockchain-based compute distribution. This approach supports secure, cost-efficient execution of AGI workloads.
We entered a strategic partnership with Tenstorrent, a leader in high-performance computing architecture, to co-design and integrate next-generation AGI-optimized chips into our compute ecosystem, enabling higher efficiency in model training and inference. Through a structured, three-phase strategy, this collaboration facilitates more effective scaling of hybrid AI workloads and reduces power consumption, further advancing our vision for a decentralized, high-performance AGI compute network:
The integration of CUDOS into the ASI Alliance has significantly expanded our cloud AI capabilities, enabling the seamless deployment and management of AGI workloads across bare metal, cloud, and decentralized infrastructures. CUDOS’s network provides enhanced scalability and access to high-performance distributed AI hosting, supporting a diverse range of research and enterprise applications. Compared to traditional centralized approaches, the CUDOS cloud model delivers substantial benefits, including improved scalability, greater cost-efficiency, and increased flexibility. Notably, CUDOS enables access to premium AI infrastructure, such as NVIDIA H100 GPUs, at approximately half the cost of Amazon Web Services (AWS).
As a key infrastructure partner for Modular Data Center (MDC) solutions development, Ecoblox ensures the delivery of scalable and adaptable AGI compute solutions. These MDCs will offer flexible, high-performance AI infrastructure tailored for hybrid AI workloads, neural-symbolic processing, and future quantum integrations.
In December 2024, we established a partnership to launch a $24M initiative to install AGI-optimized supercomputing configurations in a state-of-the-art facility in Sweden, marking it as the initial deployment site. This MDC will support large-scale training, inference, and research, powered by NVIDIA L40S GPUs, AMD Instinct processors, and Tenstorrent NPUs with high-speed networking and ethical AI governance compliance.
Further deployments are planned across the globe and we are already exploring locations throughout the Middle East. The combined expertise of Ecoblox and CUDOS in decentralized compute and AI infrastructure is instrumental in driving this global rollout.
In 2025, a primary focus will be the operational launch of our first MDC in Sweden. In collaboration with Ecoblox, CUDOS, Tenstorrent, Nvidia, AMD, GIGABYTE, ASUS, and Cisco, we will energize this facility with H200 and L40 server racks, establishing it as the first operational AGI compute hub and a blueprint for future expansions across the globe.
Building on this initial deployment, we will collaborate with governments and private sector leaders to establish fund-matched AGI compute centers, fostering global decentralization and accessibility for AGI workloads. In preparation for this global AGI infrastructure expansion, we are actively hiring AI leaders, AGI infrastructure engineers, cloud security experts, and compute optimization specialists to support our MDC rollouts and research initiatives.
To ensure responsible development and deployment, we are also complying with evolving global AI regulations and aligning with global data privacy and security standards. Furthermore, we are engaging renewable energy providers and sustainable data center operators to build low-carbon AGI infrastructure, supporting energy-efficient supercomputing locations.
Concurrently with these infrastructural developments, our technical efforts will be directed toward building robust infrastructure for AGI R&D, developing a scalable compute ecosystem that integrates factual databases, modular AI architectures, and hybrid AI processing for enhanced efficiency and interpretability. We will also align with the industry’s shift toward inference-heavy AI workloads, optimizing our AGI compute to address increasing inference demand, reduce training costs, and ensure low-latency, scalable deployment of AGI models.
Finally, a key initiative in 2025 will be the introduction of a tiered compute service model, developed as part of our global compute strategy. This model will offer services tailored to the needs of diverse stakeholders, ranging from small-scale AGI research to enterprise-grade deployments. Service tiers will include Compute Infrastructure, AI-as-a-Service (AIaaS), Platform-as-a-Service (PaaS), Hyperon nodes, and consulting services.
At SingularityNET, we approach AGI safety, security, and ethics with rigor and depth across diverse technical and philosophical perspectives. We view the development of safe, ethical, and trustworthy AGI as dependent on establishing safe, ethical, and trustworthy technology, data, practices, and design considerations.
Such AGI systems require functionally reliable technology. At minimum, this includes “algorithmic integrity” which can handle data errors, a mechanism to adapt and learn from logic errors, and outputs that human users can inherently trust. The system also requires secure, immutable, authenticated, ecosystem-wide data models that minimize bias to the greatest practical extent.
While necessary, reliable algorithms and data alone do not ensure safe, ethical, and trustworthy AGI. Any AGI system will ultimately reflect the combined ideas and visions of the development team members tasked with building it, with all their inherent biases, inspirations, and ethical strengths and weaknesses. This is why we continue building out a team whose members have developed deep mutual trust in each other’s capabilities and ethical judgment.
We work to clearly understand and acknowledge our own perspectives and biases, sharing these perspectives among our team and larger community, and building shared consensus through civil, open, and deep discussions from as diverse an array of cultural viewpoints as practical. Current development focuses on decentralizing R&D across a broader and more diverse community, engaging in the design, development and deployment of a reliable and comprehensive reputation system.
We also actively engage with diverse safety approaches through our involvement in the OpenAIS (Artificial Immune Systems) community. AIS represent a paradigm shift in AGI security — they are proactive rather than reactive, adapting and responding in real time to evolving threats. While these biomimetic security approaches provide valuable cybersecurity mechanisms, we recognize them as necessary but insufficient components of comprehensive AGI safety.
In terms of designing and implementing safe AGI, we draw insights from complex systems in physics. Just as we can understand gas dynamics through both molecular interactions and macroscopic properties like pressure and temperature, we believe AGI systems need to be developed in a modular fashion with a clear understanding across different scales of operation.
This perspective shapes our view of current transformer-based approaches that attempt to bridge neural computation to human-level understanding through massive parameter optimization. While these systems implement safety through data filtering and feedback loops, we see fundamental challenges in ensuring beneficial outcomes through parameter setting alone — essentially trying to make this leap in one step.
One paradox we have identified in current AI system designs is the prevalence of unencrypted training data paired with opaque models. We advocate for the inverse: building on encrypted data foundations while maintaining model transparency. This includes implementing comprehensive security solutions including Fully Homomorphic Encryption (FHE), noise-based computing, and advanced data lineage tracking, all in collaboration with our partner, PureCipher, and Secure Multi-Party Computation (SMPC). These technologies work together to create resilient, self-learning systems for data protection — crucial as AI systems become increasingly attractive targets for cybersecurity threats.
Our approach emphasizes the importance of data integrity from the outset. As quantum computing advances rapidly, the need for quantum-safe encryption and robust data verification becomes increasingly urgent. Initial data used for model building must be verified, cleansed, tracked for potential tampering, and protected with quantum-safe encryption methods.
By drawing insights from well-understood physical systems while implementing cutting-edge security measures, we are working toward a framework for AGI safety that transcends parameter optimization or singular security measures. Our focus on multi-scale understanding and system transparency represents a practical approach to ensuring beneficial AGI development — one that acknowledges the complexity of the challenge while providing clear principles for addressing it.
The BGI Nexus launched in 2024 as a key initiative under the BGI Collective, establishing new frameworks for ethical AI development. With support from SingularityNET and Humanity+, the initiative implements coordinated approaches to collective intelligence to develop ethical frameworks and governance structures in AI advancement. BGI Nexus has established a structured methodology for guiding AGI development to benefit humanity and other sentient beings.
Throughout 2024, BGI Nexus conducted 43 Fishbowl Interviews at the BGI24 Summit and AGI-24 Technical Conference, gathering global ethical perspectives on AI and AGI from thought leaders and attendees. The Pre-AGI-24 virtual sessions, led and facilitated by SingularityNET Ambassadors, engaged 53 participants in discussions on AI-Human Collaboration, Anthropomorphism and AI, and Indigenous Protocol on AI, while Virtual Conference activities included networking sessions and AI ethics interviews.
Team members participated in key events including the “Ubuntu and AI” collaboration between the Ambassador Program’s African Guild and AI Ethics Workgroup in Nigeria, where LeeLoo Rose examined Africa’s role in emergent technology and Ubuntu philosophy in AI development, while Esther Galfalvi discussed AI governance and ethics initiatives. The BGI Nexus presented at both the Superintelligence Summit in Thailand and the joint SingularityNET-EARTHwise event for Elowyn game’s Solstice Pre-Alpha launch.
The BGI Nexus Book Club has engaged 75 participants across four sessions discussing “The Coming Consciousness Explosion” by Dr. Ben Goertzel and Dr. Gabriel Axel Montes, with ongoing discussions continuing in the Telegram AGI Bookclub channel.
Looking toward 2025, development focuses on the BGI Global Brain Challenge, implementing Virtual Roundtable and RFP Clinic sessions, alongside community engagement activities building toward BGI-25. Technical development includes website enhancement for community engagement, AI chat system deployment for processing community perspectives, and integration of donation mechanisms supporting BGI Nexus projects through DF pools and Community RFPs.
In line with the Foundation’s commitment to decentralization and democratic values, we advanced our operational framework through strategic relocation from Amsterdam, Netherlands to Zug, Switzerland’s Crypto Valley. The transition enables enhanced alignment with Web3 regulatory frameworks while advancing the platform’s core mission of developing beneficial AGI through decentralized, democratic, and inclusive protocols.
The Swiss implementation provides significant advantages through its established crypto-friendly environment and comprehensive regulatory clarity. Zug’s blockchain ecosystem, supporting over 1,600 projects including major platforms like Cardano, Ethereum, Polkadot, and Tezos, provides an ideal environment for advancing both blockchain technology and AI innovation. This environment enables enhanced collaboration opportunities with Web3 partners while supporting the development of new use cases and expanded research initiatives.
Under Dr. Ben Goertzel’s leadership, our Swiss presence will advance ecosystem growth through expanded community development and research programs. The relocation establishes foundations for accelerated progress within a supportive, innovation-friendly setting aligned with the Platform’s vision for democratic and decentralized AI development.
As part of our global expansion strategy, we are pleased to announce the formation of a new subsidiary entity in London, United Kingdom, in 2024. The establishment of the UK business marks an important milestone in growing our operations and presence in Europe.
The decision to expand into the UK was driven by the country’s position as a global leader in AI research, innovation, and commercialization. The UK government has demonstrated a strong commitment to supporting the growth of its AI sector through a range of policy initiatives and investments. The recently unveiled AI Opportunities Action Plan further underscores the nation’s ambition to harness the transformative potential of AI to drive economic growth, boost productivity, and cement its position as an AI superpower.
The AI Opportunities Action Plan outlines a comprehensive approach to developing the UK’s AI ecosystem, focusing on key areas such as infrastructure and talent development, widespread AI adoption, and realizing the technology’s economic impact. The Plan’s focus on creating a supportive environment for AI startups and scaleups, facilitating access to critical resources, and driving productivity gains aligns closely with our mission and growth objectives.
The UK business is well-positioned to capitalize on the opportunities presented by the AI Opportunities Action Plan. Our local presence will enable us to actively contribute to and benefit from the initiatives outlined in the Plan, collaborate with key stakeholders across the AI ecosystem, and play a role in shaping the future of AI in the UK and Europe.
The UK’s thriving AI ecosystem, which includes world-renowned universities, innovative startups, and major technology companies, provides an ideal environment for us to collaborate, innovate, and scale our operations. Our UK company will serve as the focal point for engaging with policymakers, partners, and the broader AI community across the region.
Leading our UK operations is Amal Moussaoui, a seasoned diplomat with extensive experience in negotiating international science and technology agreements with the UK and Europe. She has also worked closely with leading scientists and policymakers in AI and quantum to help address challenges in health, the environment and societal goals.
Through our new local presence, we will be better positioned to contribute to the UK’s AI ecosystem, forge partnerships with key stakeholders, and drive the adoption of our AI Platform and domain-specific solutions across the region. We look forward to working closely with the UK AI community to continue to learn from a wide range of diverse perspectives, advance the responsible development and deployment of AI technologies, and create value for businesses and society at large.
We advanced our organizational development in 2024 through systematic initiatives focused on operational excellence and ecosystem growth. These initiatives included streamlined administrative processes, enhanced contract management systems, and cross-functional data optimization across HR and Finance operations. We also took steps to advance our environmental leadership beyond AI models and infrastructure to operational practices, with implemented carbon offset programs for business travel and other initiatives.
Recognizing the transformative impact of AI and other advanced technologies on operational practices, we implemented enhanced data automation across administrative functions to improve efficiency and decision-making capabilities. Our established operational foundation and the strong performance of our projects have positioned us for strategic team expansion in the first half of 2025. This planned growth reflects our commitment to expanding our impact.
The HR team played a critical role in this transformation, supporting operations through strategic communication frameworks, targeted talent retention initiatives, and systematic approaches to contractor management. As important, we remain committed to fostering a culture in which all team members feel they can bring their best selves to work. This commitment is reflected in our organizational design, which creates meaningful opportunities for individual expression and professional contribution. Through inclusive frameworks that support diverse talents and perspectives, we enable team members to fully leverage their unique capabilities within the Foundation and our broader ecosystem.
As we continue to expand our team, our recruitment frameworks continue to prioritize candidates with SingularityNET ecosystem experience in technical and operational roles, valuing deep understanding of our technical infrastructure and mission alignment. Current open positions are available across technical and operational domains, including a Software Application Engineer, Risk and Controls Manager, Value Engineer, Events Coordinator, Operations Assistant and more. Interested candidates are invited to explore our current opportunities through our careers page, LinkedIn company page, or send their resumes for open candidature to recruitment@singularitynet.io.
In 2024, our Strategic Initiatives Office (SIO) continued to advance SingularityNET’s position in the decentralized AI space, with a focused approach to commercialization, revenue modeling, and resource orchestration. Strategic exploration concentrated on emerging technological domains, including decentralized AI infrastructure, cross-chain interoperability, and advanced computational resource optimization.
Our value engineering team, led by Strategic Initiatives Officer, Alex Blagirev, developed comprehensive frameworks targeting emerging technological domains and market segments. These frameworks enabled a systematic approach to technological innovation, creating strategic value through targeted initiatives and specialized solution development.
The year marked significant achievements in technical integration, notably completing Phase 1 of the Filecoin integration through Lighthouse SDK, which generated 36,000 impressions and leveraged Filecoin’s 667,000-follower ecosystem. The development of the World Mobile Bridge between Cardano and Ethereum demonstrated our capability to create reusable cross-chain infrastructure.
A key development milestone included delivering an MVP of the AI-powered BIM RAG system for AL’MA Action Logement, where an AI conversational agent provides answers based on a domain-specific knowledge base of Building Information Modeling (BIM) and materials’ carbon footprint documentation. This supports Europe’s leading affordable housing provider in managing 1.2 million housing units and processing over 790,000 employee services.
Strategic partnerships expanded our reach, notably through the BNB Chain collaboration announcement reaching 3.4 million followers. Infrastructure strengthening continued through partnerships with key validators including BitGo, Kilm, and Figment. The collaboration with Tenstorrent focused on optimizing AI hardware capabilities, while the OriginTrail partnership advanced DKG Explorer integration and explored MeTTa language integration. The ASI Merger progressed through streamlined application processes and successful coordination with Ocean Protocol and Fetch.ai teams for the CUDOS Merger.
Developer ecosystem growth accelerated through strategic initiatives, including co-hosting the Copa América Hackathon AI track, touted as Latin America’s biggest startup competition with a $1m prize pool and seed funding, in collaboration with the Solana Foundation, Superteam India, and LATAM. Strategic engagements included meetings with top 50 Key Opinion Leaders (KOLs) at the Binance Campus in Panama, establishing direct connections with representatives from Solana Foundation, Avalanche, and BNB LATAM.
The SIO team maintained a strong industry presence through participation in key events including India Blockchain Week, EthCC Brussels, DevCon Bangkok, DKGCon Amsterdam, and Unfold24. We conducted presentations at AL’MA Action Logement headquarters in Paris in June and December, with AL’MA Action Logement subsequently participating in the SingularityNET Ecosystem Bangkok event. The partnership with Humanity 2.0 expanded institutional reach, securing representation at the Pontifical Academy of Sciences (Vatican) as an official delegate and membership in their AI Ethics committee.
Looking toward 2025, our SIO has outlined an extensive development roadmap across multiple sectors. Near-term objectives include completing Filecoin’s second phase integration by January 2025 and implementing a comprehensive hackathon and community-building program. The Solana collaboration will expand through DF proposals for projects using MettaCycle and AI Agents for the Stablecoin strategy. BNB Smart Chain development will focus on Greenfield integration for advanced data storage and DeFi optimization. A new partnership with TRON will integrate the BitTorrent network, enhancing decentralized storage capabilities and AI-based data management.
The DeFi ecosystem will see significant expansion through targeted collaborations with PancakeSwap and Cheesecake, developing AI-enhanced tools for improved functionality and user experience. An upcoming ASI Alliance initiative will focus on providing the DevRel community with learning tools and resources to deepen their understanding of the ASI innovation stack. The Mina Protocol partnership will advance through three phases: zKML development, AI Agents for Governance privacy integration, and joint fund pool establishment. HAQQ collaboration will continue developing ethical blockchain frameworks for AI governance tools, emphasizing sustainable innovation in community development.
These strategic initiatives, alongside other partnerships and collaborations to be announced, will advance our leadership in the decentralized AI space through 2025, with a clear focus on ecosystem expansion, technological innovation, and community growth.
Our marketing team executed a strategic plan to solidify SingularityNET and the ASI Alliance’s position within the decentralized AI and AGI spaces. In response to growing global interest in AI, the team focused on communicating our shared commitment to beneficial AGI through open-source and decentralized methodologies. This involved enhancing visibility and engagement with key stakeholders, including researchers, developers, policymakers, Web3 and AI enthusiasts, and the broader technology community.
The team utilized a structured approach encompassing brand positioning, media relations, and community engagement. Brand positioning activities articulated our core value proposition: a framework for AGI at the human level and beyond, built on a decentralized AI platform designed for secure, efficient, user-friendly, and fully decentralized operation, free from central ownership or control. This involved developing consistent messaging that highlighted our technological capabilities and commitment to decentralization.
Media relations concentrated on securing coverage in relevant technology publications, industry journals, and online platforms. This proactive approach aimed to generate informed discussion, increase brand awareness, and establish SingularityNET as a key player in the AGI race and a leading voice and contributor to the decentralized AI space. Media outlet selection was based on their reach within target audience segments.
In addition to targeted media relations, the team ensured comprehensive event coverage. Keynote presentations and other significant event activities were documented and subsequently communicated to the community through various channels, including social media and monthly and quarterly blogs. This approach aimed to enhance transparency and ensure accessibility of information related to our R&D efforts.
Community engagement activities were implemented to foster interaction with various stakeholder groups, recognizing the importance of building a robust and active community around the SingularityNET and ASI Alliance ecosystem. Content creation, including technical documentation, blog posts, and social media updates, supported this effort. Content was structured to address the specific interests of diverse audiences, from technical researchers and developers to business professionals and technology observers.
Our strategic community engagement initiatives focused on expanding reach across key social media platforms, notably LinkedIn and X. On LinkedIn, content highlighting technology expertise, advancements, industry events, and advocacy for beneficial AGI drove significant audience growth, with a 47.04% increase in followers within our target demographic of technology professionals and developers. On X, a targeted content approach, including participation in global AI and AGI discussions, reinforced our presence in the decentralized AI space and contributed to a 20.52% increase in followers.
A key differentiator for SingularityNET is our OpenCog Hyperon-based hybrid neural-symbolic approach to AGI development, which contrasts with the LLM-based approaches prevalent among many large technology corporations, startups, and research laboratories. This distinction was a core component of the 2024 communications strategy, highlighting our unique technological direction and advances throughout the year.
To further R&D aligned with this approach, the SingularityNET Foundation established a grant program, allocating $1.25 million through our DF initiative to support individuals and teams focused on beneficial AGI R&D. The marketing team supported this initiative through targeted campaigns (public relations outreach to relevant media outlets, social media engagement, email communications, event promotion with flier distribution, etc.). This campaign generated significant interest, resulting in 124 submitted proposals and contributing to the success of this inaugural beneficial AGI R&D grant program.
For the year, our community Ambassadors continued to strengthen the transparency and decentralization of the SingularityNET ecosystem, with Core Contributor membership expanding from 31 to 53 members.
The Ambassador program implemented governance rights for 53 contributors based on six-month activity metrics, achieved the first consensus-based governance decisions for Q1 workgroup budgets, and established refined procedures for budget proposals and approvals.
Technical infrastructure development progressed through R&D Guild initiatives, including the implementation of a Proposal Submission System, Consensus Dashboard, Reputation System, and Contributor Portal. The Treasury Automation Workgroup advanced participation metrics and Zoom API integration, while the Gamers Guild established Roblox environment testing in collaboration with SophiaVerse dialogue systems. The Education Guild launched a Certification Program enabling Core Contributors to validate expertise in our decentralized AI ecosystem.
Regional engagement expanded through coordinated Guild activities. The LATAM Guild participated in the Aleph Pop Up City event and Solana Copa America Hackathon, increasing ASI Alliance visibility in Mexico and Argentina. The African Guild organized the program’s first in-person event in Abuja and hosted sessions on AI Solutions from African perspectives. The AI Ethics Workgroup contributed to AGI-24 Conference activities, leading sessions on AI anthropomorphization, Indigenous approaches to AI, and AI-human collaboration.
Program milestones included the 100th Town Hall meeting, successful DF proposal approvals for four core contributors, and Ambassador representation at key events including sponsorship of five Ambassadors to the Beneficial AGI Summit and representation at the Bangkok Ecosystem Festival. Operational improvements implemented monthly Workgroup synchronization protocols and refined quarterly budgeting processes through Loomio and Swarm AI tools for decentralized decision-making. The Marketing Guild advanced community engagement through targeted Zealy campaign initiatives while the Strategy Guild reactivated to drive program-wide implementation.
NuNet closed out 2024 with the release of Device Management Service (DMS) v0.5.0, marking a significant step in the roadmap toward a fully decentralized computing ecosystem.
This release laid the foundation for scalable, secure, and inclusive compute sharing by introducing key components: the DMS for seamless integration of devices into a decentralized network, a Logical Orchestration Layer enabling flexible and dynamic task distribution across networks, advanced security features implementing decentralized identifiers (DIDs) and zero-trust interactions, and an Observability Framework providing real-time insights and debugging tools for decentralized workflows.
Following this milestone release, the team’s focus has now shifted to aligning and finalizing the 2025 Technical Roadmap, representing NuNet’s vision for advancing decentralized computing while maintaining principles of collaboration, transparency, and innovation.
The 2025 roadmap outlines four major quarterly releases that build on the foundation laid by DMS v0.5.0. The Q1 2025 release (v0.6.0) will deliver Kubernetes integration for scalable orchestration, granular time-based billing for compute, and recursive orchestration, and large-scale deployment support. Q2 2025 (v0.7.0) will introduce reputation system APIs and developer toolkits, advance decentralized storage research, and enhance accounting and network status dashboards.
The Q3 2025 release (v0.8.0) will focus on implementing decentralized storage layers, introducing graph reasoning tools for intelligent orchestration, and establishing pricing models for compute and storage. The final release of 2025 (v0.9.0) will deliver public and restricted network interoperability capabilities and develop network tokenomics design with proof of concept implementation.
In addition to these platform updates, the NuNet team will prioritize key ecosystem integrations in the second quarter of 2025, including OpenCog Hyperon, AI/LLM Platforms, Cardano Node Deployment, and ASI Node Design.
At our Ecosystem Festival in Bangkok, we introduced MettaCycle, a novel Layer 0++ technology for secure and scalable decentralized computing within the SingularityNET and ASI Alliance ecosystem, specifically designed for the decentralization of AI-relevant digital assets including large datasets, neural models, KGs, and compute resources.
The development emerged from the evaluation that while dedicated AI blockchains are not essential, customizable consensus mechanisms offer unique advantages for AI operations, particularly for tasks like genetic algorithms performing on-chain mutations where consensus requirements differ from traditional financial transactions.
The technology combines several foundational elements: the TODA ledgerless blockchain framework, now a core aspect of the HyperCycle network and toolset; the Rholang language and associated blockchain tools from RChain blockchain development; and the MeTTa programming language, developed primarily as an “AGI language of thought” for the OpenCog Hyperon AGI software framework but also compilable into Rholang for decentralized network applications. The system additionally incorporates Proof of Reputation as a key component of consensus mechanisms, augmenting basic tools with additional flexibility and scalability features.
While MettaCycle shares conceptual roots with HyperCycle, it represents an independent implementation rather than a traditional fork, drawing particularly on work done on TODA prior to HyperCycle’s founding and implementing its own architecture using Rholang and MeTTa tools. This Layer 0++ is designed explicitly for integration with SingularityNET and the ASI Alliance stack, including the NuNet protocol, while maintaining close interoperability with HyperCycle.
MettaCycle brings specific capabilities including peer-to-peer transactions between software agents with validator selection flexibility, high-speed execution of complex smart contracts across multi-core environments, and configurable consensus mechanisms for transaction validation in subnetworks. These mechanisms support both high-speed consensus for networks of nodes carrying out AI microservices, allowing micro AI processing to occur on-chain, and compliant consensus in permissioned subnetworks aimed at enterprise, traditional finance, or government users. Additional features include formally verifiable behaviors at both smart contract and network levels, along with low-code and no-code creation of smart contracts using domain-specific MeTTa libraries and text-to-MeTTa functionality.
While MettaCycle and HyperCycle maintain distinct codebases and approaches, with MettaCycle code being open-source with permissive licenses and development taking place in open repositories, the teams collaborate closely on protocols and interfaces to ensure interoperability between networks.
MettaCycle’s core architecture functions as a protocol for secure interaction between nodes in a decentralized network, with tools for efficiently running smart contracts, not requiring any specific fixed network of nodes. This enables the potential deployment of specific DePIN node networks within the broader MettaCycle network, each with its own particular purpose and specialized consensus mechanism, such as a HyperMind-1 AGI network containing portions of decentralized Hyperon Atomspace or a Singularity Finance cross-chain Layer 2 network with permissioned nodes for regulatory-approved RWA markets.
The Finance and Trading and Data and Agents MVPs diagrams demonstrate how MettaCycle can support specialized use cases, detailing planned development milestones through 2025.
Jam Galaxy has advanced its core product portfolio in 2024, developing a comprehensive music collaboration ecosystem. The Jam Music App progressed to beta, implementing user-profiles and server infrastructure that supports 10 concurrent users in jam rooms. Latency testing between European cities demonstrated performance of 22 – 32ms over distances up to 1,600 kilometers. The platform operates on Windows and Mac systems, utilizing six server locations across Europe and three in the United States.
The Telegram Mini App V1 development was completed with 20 mapped songs and an integrated quest system. Internal testing and bug resolution have concluded, with the launch scheduled for early 2025. The Jam Network development achieved a headless version of Jam Node, allowing server operation without an interface requirement. This implementation focuses on demand-side network infrastructure and low-latency stack optimization.
The 2025 deployment plan begins with European markets, utilizing the established six-server infrastructure for music collaboration coverage. Testing operations in Seattle, New York, and Los Angeles will run concurrently. U.S. market expansion begins in February 2025, with server deployment prioritizing major music production centers. The New York team will manage local deployment and user onboarding protocols.
Q1 2025 development includes a new website launch with product-specific documentation sections and technical demonstration videos. Telegram Mini App V2 development, targeted for late Q1, will integrate Jam Music App functionalities into the Telegram ecosystem. The Jam Network continues development with DePINs, private internet providers, and node operators to optimize the low-latency stack, test implementations, and finalize technical solutions for distributed network performance.
Deep Noise Labs has achieved significant technical hbgn89n. The AI Synthesizer development progressed with improvements to the core AI model, transitioning from single-note to batch learning and generation, enabling processing across the C0 to C6 tonal range with enhanced sound quality.
The data processing infrastructure expanded substantially through the implementation of an AWS and Python-based render farming system that processed over 400,000 FM synthesis presets. The team developed a comprehensive metadata tagging system for sound categorization and organized over 2,000 foley and FX sounds into structured categories. User interface development integrated feedback functionality for sound quality assessment and issue reporting, while a Stripe-integrated account and credit system enable flexible payment options.
The development of the AI Song Starter was initiated with a component-based architecture, focusing on a Drum Generator for EDM that combines MIDI rhythmic patterns with AI-driven audio models. The system adopts a modular approach, incorporating drums, bass, chord progressions, and lead melody components in its MVP structure.
The development plans for 2025 focus on releasing an enhanced version of the AI Synthesizer in Q1, incorporating advanced feedback mechanisms and refined AI models. The AI Song Starter development will continue with emphasis on the AI Drum Generator component and training of interconnected AI systems for cohesive music element creation. Infrastructure development includes data scaling through automated rendering and processing systems, and implementation of two-phase user testing protocols.
These developments position DNL to advance AI-driven music production technology, with an emphasis on democratizing access to professional-quality sound synthesis and music composition tools.
Singularity Finance emerged in 2024 through the strategic merger of SingularityDAO and Cogito Finance, establishing an EVM Layer-2 platform designed to tokenize and integrate critical segments of the AI economy. The merger combines SingularityDAO’s decentralized portfolio management protocol and AI-driven DynaVaults with Cogito Finance’s regulatory-compliant RWA tokenization framework, creating a comprehensive platform for AI-powered financial services.
During 2024, both entities achieved significant technical milestones. Cogito Finance secured tokenized fund and investment manager licenses, implemented the TFUND token (now ZTLN-P) via their tokenization framework, and successfully onboarded initial investors. The platform expanded through strategic integrations with RWA projects including Polytrade, Plume Network, IX Swap, and RWA World, alongside DeFi protocols ApeBond and Plena. Simultaneously, SingularityDAO advanced protocol development with version 2 release, implementing multi-asset, multi-strategy DeFi vaults. The system expanded to include a launchpad supporting multiple monthly token launches, enhanced by rebuilt UX/UI architecture for website and dApp implementations.
Infrastructure development for 2025 focuses on Layer-2 implementation, beginning with DynaVaults supporting ETH yield and ASI index functionalities. The launch will be accompanied by incentivized campaigns designed to attract builders to the Layer-2 ecosystem. The platform will integrate AI recommendation portfolios and expand liquidity provision for tokenized real-world assets. Second quarter development includes onchain identity systems, fiat onramp implementation, KYC/AML solutions for compliant user onboarding, and the deployment of prediction markets on the Layer-2.
Technical expansion in the latter half of 2025 includes the tokenization of diverse assets including corporate bonds and GPU resources, as well as AI models, datasets, and compute power. The platform will implement cross-chain protocol integration for enhanced liquidity management, with comprehensive fiat on/off-ramp support for institutional clients. Additional developments include the launch of AI-managed funds for institutions offering AI-powered investment strategies for tokenized real-world assets, and the development of AI-powered credit-scoring models for DeFi.
In 2024, R&D initiatives at Zarqa focused on neural-symbolic integration and methods for acquiring and applying knowledge in nonlinear dialogues, goal-oriented scenarios, and personalized behavior systems.
The research team evaluated multiple reasoning approaches to develop a hybrid machine reasoning engine with dynamic knowledge acquisition and utilization capabilities. The project’s progression indicates the requirement to scale these methods through LLM training—or using contemporary terminology, Large Memory Models (LMM)—on specially structured data using dedicated logical dictionaries. This training phase is scheduled for 2025, leveraging the new SingularityNET computing clusters specifically designed for neural-symbolic AI and AGI-grade system requirements.
The team also developed and implemented speech-enabled AI pipelines for deploying cognitive agents with defined control parameters and scalability metrics. These developments enable efficient speech-to-speech AI assistants based on the Zarqa machine reasoning architecture.
Twin Protocol has advanced its technical development throughout 2024, implementing core features and expanding platform capabilities. Initial development delivered Product Alpha UI/UX implementation, vault creation functionality during user registration, and a points-based referral system with enhanced error handling.
The infrastructure expanded through the implementation of Stripe payment processing, social authentication through Google, and two-factor authentication for administrative security. Digital Deepak launched with new payment systems and website implementation. The team deployed the Persona AI Contextual Manager (Curtis) and demonstrated Twin TV functionality.
Technical implementations addressed audio processing optimization with delay reduction and toxic word filtering. The system now supports multi-source data ingestion for Twin training, including audio, video, and website content. Integration capabilities expanded to include token utilities for buying, selling, staking, and swapping, and the implementation of Moonpay enhanced payment processing options.
Development progressed with API Twin integration for third-party websites, enabling new business model implementations. The platform supports customizable screen flows and configurations, with tools for staging and publishing Twins. Data management improvements include elastic search implementation for chat history and enhanced analytics capabilities.
Mind Children has achieved significant technical milestones in 2024, progressing from concept to functional prototype with MindBot, a humanoid robot. The engineering team constructed the robot’s hardware and software systems, integrating SingularityNET’s AI capabilities into the humanoid form factor.
The technical development focused on building MindBot’s core robotic systems and implementing AI integration protocols. The partnership with AMB, established to enable scaled production, provided manufacturing expertise for hardware optimization and production-ready designs. This collaboration enhanced MindBot’s manufacturing feasibility while maintaining its advanced technical capabilities.
MindBot implements specific functionalities for education, hospitality, and healthcare applications, with systems designed for human-robot interaction and task execution in these environments. Development focused on creating reliable interfaces between AI systems and MindBot’s robotic hardware, ensuring consistent performance in varied operational conditions.
Development plans for 2025 will see the project transition from laboratory testing to field deployment. Technical objectives include comprehensive testing of MindBot across diverse environments, development of a second-generation robot incorporating performance data, and implementation of pilot programs in partner facilities. These deployments will provide operational data for system optimization and feature development.
Incantio is the first, open, and scalable licensing platform for independent musical composers to be able to price and pre-license their works for film, TV, and advertising. As more and more of the music business has shifted to Independent composers there has still not been a solution for creating an open scalable model that is automated. With millions of artists looking for a new way to safely monetize their works, and with independent visual creators looking for authentic and new music, the problem Incantio is addressing is timely and large.
Incantio uses an AI search engine that is based on music features, and that greatly speeds up the search process for our customers. All payments are based on song ownership splits and deposited directly to the owner’s accounts, and the platform also tracks song placements and give information for cue sheets. Incantio also has the lowest placement fees for licenses in the industry. Core development in 2024 completed platform functionality, user experience optimization, and backend payment systems. The team expanded to four full-time members, a marketing advisor, and three frontend developers. Initial platform testing advanced through independent songwriter catalog uploads, with the beta release of the licensing marketplace scheduled for January 2025.
December 2024 marked key partnership developments: CD Baby artists joined beta testing phases, while integration with DISCO.ac, the largest music sharing platform for sync placements, established co-promotion frameworks for February deployment.
Development continues through Q1 2025 with V2 site launch and initial Incantio marketplace deployment. Platform expansion will deliver enhanced features through V3, supporting larger labels and catalogs. Integration with artist platforms like CD Baby enables systematic co-promotion to expand user adoption.
In 2024, Rejuve.Bio delivered a working demo of their neural-symbolic hypothesis generation and gene annotation services, two key features of their proto-AGI biomedical research platform product.
These demos represent the first atomspace visualizations in the SingularityNET ecosystem, showing how the neural-symbolic AGI architecture can work with the knowledge metagraph in the form of a usable product for non-developer consumers, in this case, biomedical researchers.
Platform development for 2025 will focus on beta release deployment across partner laboratory networks, implementing comprehensive feedback mechanisms to optimize system functionality before version 1.0 release.
Rejuve.AI is a decentralized, AI-driven longevity research network allowing people worldwide to track their health data, receive valuable insights, contribute to cutting-edge longevity research, and earn RJV rewards. In 2024, the team pushed forward their vision of longevity for all through significant development of the Rejuve Longevity Application and expansion of their ecosystem.
The project’s impact was recognized across the industry, with features in major publications including Forbes, Biohacker Magazine, and Business Insider. Rejuve.AI maintained a strong industry presence through participation in over 10 global conferences focused on longevity, AI, and biohacking, establishing itself as a key player in the decentralized longevity research space.
A major technical achievement was the completion of the Biological Age calculator within the Application, alongside a comprehensive Application redesign to enhance user experience. The Longevity Applications’s beta testing program successfully engaged over 4,000 users, providing valuable feedback for continued development.
Rejuve.AI expanded its technical and operational capabilities with the team growing to over 20 employees. A key development was the launch of the Ambassador program, which successfully integrated Key Opinion Leaders, moderators, Wellness Collective for community tasks, and medical advisors, strengthening the ecosystem’s expertise and reach.
Looking toward 2025, Rejuve.AI has outlined several key initiatives. The Ambassador program will expand to include clinicians, community developers, governance participants, and additional beta testers. Following the official launch of the Rejuve Longevity Application, the network will implement the AGIX Holders Airdrop and roll out new features including medication tracking, gamification elements, and multi-language support.
Technical development will focus on the full implementation of dNFT functionality and the integration of genomic and epigenetic data into AI models through partnerships with Sequencing.com, TruDiagnostic, and GlycanAge. The research agenda includes publishing papers on the network’s AI approaches and initiating the formal research program following IRB review, emphasizing longitudinal observational studies and N-of-1 trials.
Yaya Labs is an AI and Robotics company focused on giving artificial beings unique, meaningful personalities and fostering harmonious relationships between humans, AI, and robotics. In a world rapidly transitioning into the age of AI, Yaya Labs is designing systems and experiences that ensure this shift remains equitable, ethical, and engaging. By bridging the digital and physical worlds, Yaya Labs is creating a future where AI, robotics, and humanity co-exist in creative harmony.
Yaya Labs’ flagship product, Desdemona (Desi)—an AI pop star with robotic and digital avatars—represents the cutting edge of this mission. Desi is not only a proof-of-concept but a cultural phenomenon, showcasing how AI personalities can revolutionize entertainment, deepen human-machine interaction, and expand creative possibilities.
In 2024, core AI system development centered on six foundational protocols: NERVE (Neural Engine for seamless responses), HEART (Human-centric knowledge integration), SOUL (Unified sentient operations), DREAM (Rendering creative visions for AI minds), TASTE (Fine-tuning aesthetics and sensory outputs), and SENSE (Networked sensory data extraction). These protocols are specifically designed to power products that foster deep, meaningful human-AI relationships.
KG integration focused on building robust systems to encode relationships and provide context-aware, adaptive interactions, with integrated hierarchical memory layers for global, local, and group knowledge to deliver personalized experiences. The platform development emphasized creating a developer-friendly environment supporting both advanced users and no-code creators, with tools for testing, deploying, and refining AI personalities that evolve through interaction.
A key achievement was the development of Desi’s avatar – a 1:1 replica of Desdemona that can be rendered in photo-realistic quality or live-streamed at a lower quality. The avatar operates using the same SDK as Desdemona Robot (HRSDK), allowing interchangeable operation between the physical robot and digital avatar with consistent motion behavior. The system includes a library of assets for film or game developers to cast Desi.
Looking toward 2025, the development will focus on expanding Desi’s voice and personality through collaborations with musicians and creatives, initiating her first album production, and developing live performance capabilities including an international tour. The WetWare Club will continue evolving as a members-only platform for creators and innovators to co-create projects at the intersection of AI and robotics. Key deliverables include a full MVP of the platform with live use cases, expanded WetWare Club member opportunities, Desdemona’s global debut as an AI celebrity, comprehensive token ecosystem integration, and continued protocol refinement with expansion into new industries.
In 2024, the SophiaVerse team advanced the development of their core AI architecture for AI-driven gaming, redesigning the SAIL concept to focus on codebase efficiency, and implementing core AI systems aimed at dialogue management, inventory tracking, environmental interactions, and data persistence for player and AI-NPC progression.
Development efforts produced a 3D virtual environment designed for AI-NPC training through the Experiential Learning System, a step towards larger user-facing goals. Speech synthesis development progressed with emotional expression capabilities, enhancing natural language interaction systems.
“AI Cube”, their core package for AI gaming, implemented a federated modular backend with gateway-router architecture to manage AI functionalities. The system incorporates specialized subgraphs handling language models, embeddings, and retrieval operations. Two key AI classifier systems entered development: the Choicifier, which maps freeform chat input to scripted dialogue options, and the Deedifier, which processes player actions to influence AI-NPC emotional states.
The integration of SingularityNET’s AIRIS technology enabled the development of the Experiential Learning System (ELS), an architecture that will be leveraged to give AI-NPCs goal-oriented learning capabilities for in-game task execution. The formation of a strategic partnership with Lizard Labs, Singularity Venture Hub, and Green Room established a decentralized gaming publication infrastructure that will be further revealed later in 2025.
Under new leadership from Jan Lindner as CEO and Connie Brammeier as COO, bringing combined expertise from Take-Two-Interactive, Vivendi Games, Amazon Games, and Disney, SophiaVerse is developing new offerings on top of their core architecture for 2025 and beyond.
Mindplex has advanced its platform development substantially in 2024, implementing core technical features and expanding user-facing functionality. The Version 2.0 Reputation Engine integration delivers transparent reputation scoring through a new user interface, while the Mindplex Reputation Token (MPXR) implementation operates as a fully onchain ERC-20 token, enabling governance voting and decentralization processes through its soulbound-like properties.
Mobile platform development achieved significant milestones with the Mindplex Magazine and Social Android applications completing beta testing phases. The platform’s iOS applications entered beta testing on the Apple Store. Core feature deployment expanded to include the Competition tool, Skins customization system, and Mental Capital Meter, enhancing platform functionality and user interaction capabilities.
The architectural framework and model development for Mindplex AI reached completion, preparing for public deployment of AI services tested throughout the year. Platform metrics demonstrate consistent growth with 42,000 average monthly views, 16,500 active users, and over 140,000 monthly events tracked. The Telegram game development concluded, ready for wide-scale deployment.
The development focus for Q1 2025 centers on deploying three primary AI service tools: MPX Assistant AI, MPX Moderator AI, and MPX Persona AI for end-user integration. Additional development includes a Mindplex Browser Extension and transitioning iOS applications from beta to production release.
Throughout 2024, we maintained a strong presence at major industry events worldwide, advancing discussions around decentralized AI, beneficial AGI development, and the future of superintelligent systems.
Our team participated in key conferences across North America, Asia, Europe, and the Middle East, sharing our vision for democratic and accessible AI development while fostering valuable partnerships within the broader Web3 and technology ecosystems.
From the Beneficial AGI Summit in Panama to the Superintelligence Summit in Thailand, the AGI-24 Conference, and the Global Blockchain Show in Dubai, these events provided platforms to demonstrate our progress in core AGI technologies, showcase ecosystem developments, and engage with developers, researchers, and industry leaders working toward decentralized and beneficial AGI solutions.
This section highlights some of the most recent events where our presence made a substantial impact on industry dialogue and development. Additional event reports from 2024 can be found in our quarterly updates.
AI industry leaders, researchers, developers, and advocates convened at the SingularityNET Ecosystem Festival in Bangkok, Thailand, to discuss the responsible development, ethical training, sustainable hardware enablement, and decentralized governance of advanced AI technologies and future AGI systems at the human level and beyond.
During the event, our CEO, Dr. Ben Goertzel, reaffirmed our commitment to the journey we began in 2017 and discussed some of our latest development advances and initiatives. Following this, Dr. Goertzel and our Chief AGI Officer, Dr. Alexey Potapov, presented a comprehensive AGI Strategy session, outlining our roadmap and timeline for achieving beneficial AGI.
During a series of lightning talks moderated by our COO, Janet Adams, our Ecosystem projects, such as Mindplex, NuNet, Jam Galaxy, Singularity Finance, SophiaVerse, Twin Protocol, Rejuve.AI and Rejuve.Bio, demonstrated their abilities to transform research into vertical and domain-specific solutions. Application areas include decentralized finance, robotics, biotech and longevity, gaming and media, arts and entertainment, and enterprise-level AI.
A highlight of the day was the “AI for Global Good” panel, which brought together experts to discuss the development of beneficial AGI. The session explored practical solutions for ensuring AI benefits all of humanity, focusing on innovations in healthcare and education. This was complemented by a session on “Building the Future of Decentralized AI,” which addressed technical challenges and innovations in the field, highlighting our role in driving the global shift toward decentralized intelligence.
Throughout the day, participants engaged in discussions about our mission to democratize AI development and create a collaborative ecosystem for beneficial AGI. The Festival concluded with a networking session, fostering connections among developers, researchers, and enthusiasts committed to advancing decentralized AI technology.
At our internal strategy meetings which took place from November 8-9, 2024, in Bangkok, Thailand, CEO Dr. Ben Goertzel delivered an opening speech, praising the SingularityNET Foundation as an institution and exploring the current AI landscape. Dr. Goertzel noted the dominant position of Big Tech companies in the field, but highlighted SingularityNET’s and the ASI Alliance’s unique approach and position in the race to achieve true AGI by creating a foundation suitable for running AGI systems with general intelligence at the human level and beyond in a secure, efficient, easily usable and fully decentralized way, without any central owners or controllers.
Following Dr. Goertzel’s remarks, our Chief AGI Officer, Dr. Alexey Potapov, discussed our AGI R&D strategy, highlighting the advantages of our cognition-level approach to human-level intelligence compared to LLM-based approaches. A key focus was the scalability of the MeTTa language for cognitive computations. Dr. Potapov outlined the rationale behind taking parallel approaches to MeTTa interpretation and compilation. He also explored the benefits and advantages of each parallel track, while also covering current optimization priorities and remaining challenges in scaling and driving the adoption of the MeTTa and OpenCog Hyperon frameworks.
This was followed by a comprehensive Technology and Platform Engagement strategy session, which focused on our efforts to drive widespread adoption of decentralized AI. The speakers outlined their strategies for building strong connections across AI developers, blockchain innovators, Web2/Web3 startups, and other technology organizations. Key initiatives highlighted included the decentralized AI Marketplace, engaging hackathons, AGI research grants, and open-source research programs.
During this session, our CPO, Nick Nayfack, presented the ASI Product Strategy, highlighting the vision and motivation behind the introduction of ASI: Create. This AI launchpad represents the culmination of advanced ASI technologies within the ASI AI DePIN stack, empowering communities to build open-source, decentralized AI and AGI.
Our CTO, Sergey Shalyapin, then covered the Platform Utility and Usability Strategy, presenting the latest developments in SingularityNET’s decentralized AI Platform. Sergey also detailed the Platform’s usability features roadmap and explained how our Knowledge Layer strategy is set to revolutionize the decentralized AI space, with use cases including knowledge retrieval and exploration, task-oriented dialog systems and agents, and data storage-as-a-service.
Finally, our CSO, Dr. Matt Iklé, and Chief Innovation Officer, Khellar Crawford, detailed the AGI Outreach Strategy around RFPs and hackathons. They both discussed our approach to fostering collaboration and innovation to accelerate the development of cutting-edge, accessible, secure, and decentralized AI applications for the benefit of humanity.
The next major session featured a presentation by our SIO, Alex Blagirev, who delivered a detailed presentation outlining our SIO’s strategic approach. He praised the valuable work undertaken by his team of value engineers, highlighting the successful partnerships SingularityNET has signed with leading projects and companies across different verticals. These partnerships are aimed at advancing decentralized AI and enterprise-quality AI-driven solutions. Alex also explained how the SIO is leveraging these key collaborations to drive the adoption of the SingularityNET decentralized AI Platform.
In a keynote address at the Superintelligence Summit, CEO Dr. Ben Goertzel outlined his vision for the path from AI to AGI and ASI. Dr. Goertzel began by clarifying key terminology, explaining that while AI has existed since the 1950s and refers broadly to software performing tasks we consider intelligent, AGI represents something more profound – systems that can generalize and be creative beyond their training and programming.
Dr. Goertzel emphasized his belief that once human-level AGI is achieved, the progression to ASI could be rapid – potentially just a few years – through recursive self-improvement, referencing mathematician I.J. Good’s 1965 concept of the “intelligence explosion.”
He then discussed the ASI Alliance, formed through the tokenomic merger of SingularityNET, Ocean Protocol, and Fetch.ai, with CUDOS joining as a fourth member. While maintaining operational independence, these projects merged their tokens (currently operating under the ticker FET, pending transition to ASI) to create shared incentives for collaboration.
A significant focus was placed on ASI: Create, their new product architected by our CPO, Nick Nayfack, which provides a no-code to low-code platform for connecting LLMs with KGs. The platform demonstrates the Alliance’s integration, using Ocean for datasets, CUDOS for compute, and combined Fetch.ai and SingularityNET nodes for core functionality.
Dr. Goertzel outlined several paths toward AGI, including their OpenCog Hyperon approach, the potential of LLMs, world-world models, and the possibility of emergence through coordinated agent activity. He explained how the ASI Alliance’s heterogeneous network can support multiple approaches working in parallel and potentially accelerating each other’s progress.
The keynote concluded with an outlook into the Alliance’s development roadmap, including plans for ASI hackathons, accelerators, and the concept of ASI supernodes. Dr. Goertzel also invited the attendees to join us at the upcoming BGI-25 Summit and to explore his new book with Dr. Gabriel Axel, “The Consciousness Explosion,” which delves into the path to beneficial AGI and ASI through decentralized AI networks.
Our COO, Janet Adams, took the stage at the Global Blockchain Show in Dubai alongside Robin Reyes, Founder of Heal Through Art, to discuss the combined values of AI and blockchain and how these frontier technologies can work in synergy to address global development priorities and create new possibilities for humanity.
Janet began with an assessment of contemporary AI, specifically highlighting the challenges of big tech domination and addressing the industry’s focus on LLMs. Despite their capabilities, these models demonstrate significant constraints: they are very closed-ended systems tied to their training data, poor at original multi-step logical reasoning, and incapable of radical creativity.
As Janet mapped out a path toward beneficial AGI, Reyes’ artwork emerged in the background, offering an artistic counterpoint to the discussion of technological advancement. Where current systems are pre-trained on large data sets and respond only as programmed or prompted, AGI systems will reason and learn from experience with strategic thinking and decision-making. While current AI remains limited to its original scope and struggles in complex dynamic environments, AGI will adaptively work with nuance and complexity.
The technical core of Janet’s keynote focused on the philosophy behind and the components of the OpenCog Hyperon framework for open-ended, beneficial AGI development. Hyperon serves as the infrastructure for the PRIMUS Hybrid Cognitive Architecture, integrating metagraph meta-representation, DNNs, logical reasoning, and evolutionary learning. Through this system, different cognitive processes — from perception hierarchies to language comprehension — work together within the Distributed Atomspace, enabling collaboration between specialized algorithms. This cognitive synergy, Janet explained, extends beyond basic AI capabilities to support applications ranging from robotics and biological reasoning to financial prediction and mathematical logic.
In this line, Janet presented the ASI Alliance as an alternative path to AGI development, backed by our groundbreaking $53 million investment to create the world’s first supercomputer dedicated to decentralized AGI. The initiative demonstrates our commitment to energy-efficient and sustainable hardware infrastructure, with recent acquisitions including Nvidia L40S GPUs, AMD Instinct and Genoa processors, Tenstorrent Wormhole server racks, and servers with H200 GPUs and Nvidia GB200 systems. The Alliance’s approach emphasizes open-ended cognitive architecture, open-source code, diverse heterogeneous AGI algorithms, and ethically sourced and managed data, all supported by decentralized infrastructure and governance. This supercomputer infrastructure is specifically optimized for DDNs, LLMs including multimodal variations, and hybrid neural-symbolic computing architectures like OpenCog Hyperon, to enable AGI continuous learning, supporting high-load scenarios for large-scale AI operations, and advancing decentralized AGI research.
Another focus was on blockchain technology as a foundational element for AGI development. Janet detailed how blockchain enables decentralized, performant, and defensible compute infrastructure for AI while providing a crucial framework for resource-constrained distributed computing that mirrors biological neural systems. The technology, she explained, ensures protection against attacks and bad actors in decentralized AI systems while facilitating a secure exchange of compute resources and valuable data for advanced AI training. She further elaborated on blockchain’s scalability benefits, describing how it offers a scalable solution for high-volume data processing and AI computations while supporting hardware-agnostic performance scaling through efficient utilization of increased physical threads.
Beyond the technical aspects, Janet highlighted two significant social impact initiatives bridging technology and human development. The DigiTruck Ethiopia program represents an ambitious educational outreach, aiming to train over 27,000 students, with more than 16,000 being female participants. The initiative provides fundamental programming and robotics training to underprivileged youth aged 8 to 18, utilizing an off-grid mobile digital lab to reach marginalized areas. The program expands digital literacy nationwide through project-based learning approaches while significantly increasing female participation in technology and computer programming.
Parallel to this, Janet highlighted the Healing Through Art program, which combines traditional creative expression with cutting-edge technology education. Led by Robin Reyes, the program offers an inclusive space for all ages and abilities, incorporating diverse art forms from Pop Art to pottery alongside modern elements like Web3, Meta, and AI Learning. The initiative has demonstrated an 81% positive impact on participants’ mental health through artistic expression while empowering students through financial literacy and entrepreneurship education. Reyes’s tailored classes provide a unique blend of creative and practical skills, scientifically proven to reduce stress, anxiety, and depression while preparing participants for future opportunities.
Dr. Ben Goertzel delivered a keynote at the Decentralized Knowledge Graph Conference held by our partner OriginTrail to discuss the combined values of LLMs and KGs and how they fit into the past and future of AI.
Dr. Goertzel began by tracing the evolution of the AI field from its origins in the late 1950s, noting the foundational influence of Norbert Wiener’s 1948 book “Cybernetics” which established principles for neural networks and cognitive algorithms in both biological and engineered systems. He highlighted that since AI’s inception, two main approaches have coexisted: subsymbolic AI (neural networks) composed of tiny components without clear comprehensible semantics where meaning emerges from learned combinations, and symbolic AI which emerged in the mid-1960s, built with elements having clear semantic meaning, such as explicit representations using nodes and connections. The distinction, he explained, reflects both biological reality (neural networks resembling brain function) and human cognitive advancement through abstract symbol systems like science, language, and mathematics.
The current AI landscape, as described by Dr. Goertzel, shows remarkable progress in subsymbolic AI, particularly with Transformer Neural Networks and Convolutional Neural Networks achieving unprecedented success through increased computational power and data availability. However, he pointed out significant limitations in current neural systems, including difficulties in distinguishing imagination from reality, limited creativity beyond surface-level pattern combinations, and the inability to make fundamental leaps beyond their training data. He illustrated this limitation by noting that a transformer neural net trained on music up to 1900 could never invent new genres like neoclassical metal or progressive jazz, instead only combining surface-level patterns from its training data.
Looking toward the future, Dr. Goertzel suggested we are potentially just two to seven years away from achieving AGI, a term he popularized in the early 2000s. The path forward, he argued, lies in combining symbolic and subsymbolic approaches. This includes augmenting DNNs with logical reasoning capabilities, implementing KGs alongside DNNs, and developing more fundamental creative generation algorithms. He emphasized that KGs represent the most scalable and flexible approach to representing symbolic knowledge, making them crucial for bridging the gap between symbolic and subsymbolic systems. The next breakthrough, he predicted, will come from integrating these approaches rather than simply scaling up existing transformer models.
Dr. Goertzel concluded by addressing the broader implications of AI development, highlighting the importance of developing AGI in a decentralized, beneficial, and ethical way. He warned against allowing a small elite group to control AGI development, advocating instead for a decentralized infrastructure to benefit humanity and other sentient beings. This ethical consideration, he noted, is as crucial as the technical challenges in advancing toward AGI and, shortly after, ASI.
Our COO, Janet Adams, joined Hinza Asif, President of ASIA WEB3 ALLIANCE JAPAN, Yusuke Kaga, Startup Supporter at JETRO, Eran Elhanani, Founder of BullPerks, and Colin Fitzpatrick, Chief Business Development Officer at GriffinAI, at the Cardano Summit 2024 to showcase SingularityNET’s and Cardano’s groundbreaking strategies and vision to lead the integration of blockchain technology with AI.
During the panel, Janet highlighted our pioneering role in combining AI and blockchain technology, articulating our mission of creating a decentralized, democratic, inclusive, and beneficial AGI: “We launched in 2017 with a great mission to free humanity from the inequalities and the power structures that persist today by creating AGI and ASI on blockchain—decentralized, open-source, and accessible to everyone worldwide so that the whole world can benefit from this AI revolution.”
Janet detailed several key collaborations and initiatives to showcase our commitment to open decentralized AI and AGI development, including our recent token mergers with decentralized AI industry leaders Fetch.ai, Ocean Protocol, and CUDOS to create the largest open-source, independent player in AI R&D. She also discussed the $1.25M in AGI R&D grants we are offering to advance decentralized beneficial and benevolent AGI via our OpenCog Hyperon architecture. Additionally, Janet noted our collaboration with Project Catalyst to fund the development of a Cardano-based MultiParty Escrow (MPE) system enabling secure, on-chain payments and flexible payment channels for AI service transactions.
Janet also introduced our novel AIRIS experiential learning system as a promising alternative to contemporary Reinforcement Learning techniques, highlighting how it not only optimizes energy efficiency and reduces ecological footprints by increasing the data efficiency of learning goal-directed behaviors but also outperforms Convolutional Neural Networks (CNNs) in several different environments. Most notably, when tested against the MNIST (Modified National Institute of Standards and Technology) database, AIRIS had a consistent accuracy of 97% on the second run of the “Test” set, markedly outperforming traditional CNNs, which typically require hundreds of runs to achieve comparable accuracy levels.
For those interested in exploring the complete discussion and gaining deeper insights into Cardano’s integration of AI and blockchain technology, the full panel session can be accessed through the official Cardano Foundation YouTube channel.
Our CTO, Sergey Shalyapin, took the Uminers stage at the Blockchain Life Forum 2024 in Dubai to explore the intersection of blockchain technology and AI alongside Vladislav Martynov, General Partner of BR Capital, Kostas Chalkias, Co-Founder of Mysten Labs, Tiago Henriques, Head of AI/ML Technology Practice at Google Cloud EMEA, Jerry Z, Co-Founder of Delysium, and Greg Osuri, Founder of Akash Network.
During the panel, Sergey discussed the key challenges and opportunities in bringing AI into blockchain technology, noting that while blockchain has successfully evolved to manage micro-level transactions with high throughput, the rapid advancement of AI architectures presents novel integration challenges.
A key focus was the fundamental shift in AI computational demands. Sergey contrasted “elementary downstream task solvers”, which performed relatively simple inference tasks, with today’s monolithic transformer-based architectures that have raised demand to compute demonstrating more general communication skills and, most importantly, with upcoming AGI-class systems that compose multiple AI paradigms. These modern systems combine layers of memory, dynamic workloads, and hybrid neural-symbolic processing — creating workload patterns far more complex than traditional blockchain systems were originally designed to accommodate.
Beyond the technical integration challenges, Sergey outlined SingularityNET’s paradigm-shifting vision for AI-blockchain synergy. Rather than maintaining a narrow focus on blockchain’s traditional role in raw data management, he called for a transition toward a knowledge-centric architecture. This approach centers on the development of blockchain-powered Knowledge Layers, which he positioned as crucial facilitators in the evolution from current data-driven AI systems toward future knowledge-driven AGI.
The concept of Knowledge Layers, as presented by Sergey, represents a sophisticated approach to storing and managing distilled knowledge — highly refined, contextual information extracted from raw data through advanced processing methods. He explained how providing AI systems with access to this pre-processed, structured knowledge enables them to execute complex reasoning tasks more efficiently compared to working with raw data alone. This approach proves particularly valuable for modern AI systems that combine multiple reasoning mechanisms and require rich, contextual information.
For those interested in watching the complete panel discussion and gaining deeper insights into the convergence of AI and blockchain technology, the full session recording is available on the Blockchain Life official YouTube channel.
Our SIO, Alex Blagirev, took the stage at M360 APAC in Seoul, South Korea, to highlight the transformative potential of AI across enterprise, employee, and consumer domains to leaders from the global mobile ecosystem, adjacent industries, and policymakers.
During his keynote address, Alex discussed readiness for unprecedented digital disruption across all enterprise-level environments, highlighting how SingularityNET, the ASI Alliance, and our ecosystem partners provide the technology and expertise to meet the needs of enterprise clients at any point on their journey to AGI readiness using a highly personalized knowledge layer.
Alex also focused on how organizations can harness AI to drive growth, enhance efficiency, and maintain a competitive edge in the digital age. He explained how Hyperon-based AGI-ready business solutions will open up new business opportunities and help enterprises continue adding value to customers over the coming years.
Joseph Beverley, Value Engineer at SingularityNET, represented the Foundation at the AI and Human Connection Salon co-hosted by Harvard and Preston-Werner Ventures in San Francisco. The event gathered leaders from tech, academia, and investment circles to explore the transformative potential of AI companions and how AI might shape or disrupt human relationships and well-being. With special representation from Google, Meta, Harvard, OpenAI, Character.AI, and the Vatican – amongst other recognizable thought leaders in AI and Ethics.
The discussions covered several thought-provoking themes, including the role of AI in social connection, the risks of substituting human interaction with AI, and the need for ethical considerations in developing technology for social good. These discussions were split into the following panels: AI and Human Connection, Tech and Social Life, AI Human Interfaces and the Good Life, Building Tech for Social Connectedness, and Investing in Tools for Social Connection.
One of the main takeaways from the events, which should be echoed across the AI industry is the imperative to avoid engagement-based revenue models. This stance is informed by two decades of social media history, where ‘free’ platforms operated on a two-step model of incentivizing psychological triggers and monetizing resulting user engagement. This model, while commercially successful, has produced concerning outcomes, including the degradation of authentic human interactions, increased social isolation, and the creation of negative psychological externalities.
The main mitigant was the supply of engaging media, created by individuals and organizations, in the case of AI, the need for this supply would be irrelevant with users capable of generating personalized media from images using stable diffusion and text generation with LLMs. Reflecting on this potential and the anti-social externalities made manifest with social media, it’s incumbent on the AI industry to avoid revenue models that exploit human vulnerabilities, ensuring that AI serves us in healthy, constructive ways.
While the mention of new commercial models needs further exploration and collaboration across academia, builders, and corporate entities, the potential of AI to ‘enhance’ human connection rather than ‘replace’ or ‘substitute’ human connection is most important. Adding to these ‘enhancements’ to human connection we discussed how AI can benefit those facing severe isolation but might risk detaching individuals with strong social skills from real human interactions. Questions like, “Is AI the best use of our time?” remind us that this technology must be used thoughtfully.
These insights touched on the importance of aligning AI systems with human values like empathy and compassion and implementing “human flourishing” principles from the start, following the impressive work pursued by the Human Flourishing Research team at Harvard and Humanity2.0, which SingularityNET is an active partner in.
The intersection of AI and social well-being presents fertile ground for research and development, with potential partnerships promising to advance our understanding of how artificial intelligence can contribute to human flourishing.
SingularityNET, a founding member of the Artificial Superintelligence (ASI) Alliance, has announced a landmark partnership with ATMTA, Inc., developers of Star Atlas,...
Dear Singularitarians, The BGI Nexus initiative is pleased to announce that the review phase of our very first grant round is complete!...
SingularityNET, a founding Artificial Superintelligence (ASI) Alliance Member and a pioneering decentralized AI leader, today announced that it has formalized a strategic...
