We are excited to announce that our COO, @JanetAdamsAI, will speak at @wireditalia's THE BIG INTERVIEW #TBI25 on June 26, 2025, at Bocconi University, Milan. Learn more and register to attend: https://t.co/CoUaxwD1uA https://t.co/utLF8lx9Fu

Active Predictive Coding (ActPC) comprises a biologically inspired framework for learning and inference. By minimizing prediction errors between internal models and observed data, ActPC iteratively refines both latent representations and model parameters. Unlike backpropagation-based neural networks, ActPC emphasizes local error signals, making it inherently more suited for real-time, online learning and enabling integration of reinforcement learning and symbolic reasoning. Traditional gradient-based optimization in neural networks struggles to support large-scale real-time learning dynamics due to the brittleness of the underlying backpropagation algorithm, which requires carefully coordinated and synchronized updating across a large network (leading to a reliance on large batch-based updates), and which suffers convergence problems when neural architectures are too recurrent or otherwise too complex. These shortcomings lead to an unfortunate sociotechnical dynamic in which neural architectures oriented toward robust AGI-oriented learning, reasoning, and memory are insufficiently pursued because they tend to involve network topologies for which backpropagation will not easily converge. ActPC resolves these issues on both conceptual and mathematical levels, but can also suffer from lengthy convergence times and undesirable transient dynamics. Information-geometrically enhanced ActPC (ActPC-Geom) provides a compelling potential alternative: by incorporating measure-dependent operators derived from the Wasserstein distance, one aligns parameter updates with the natural structure of the probability distributions underlying ActPC, thereby accelerating learning and smoothing out digressive transient dynamics that might otherwise occur. Like most practical applications of information geometry, ActPC-Geom faces significant computational challenges. However, we believe these can be addressed via appropriate deployment of machine learning and reasoning algorithms within the geometric modeling itself (using neural approximators and information-geometry-guided kernel-PCA-based embeddings). Further, these ML and MR algorithms can be used to inject valuable cognitive properties into the neural learning process, alongside providing acceleration. Read the full paper to learn more: https://t.co/Q62zdVPhIu

Active Predictive Coding (ActPC) is a biologically inspired framework for learning and inference. By minimizing prediction errors between internal models and observed data, ActPC iteratively refines both latent representations and model parameters. Unlike backpropagation-based neural networks, ActPC emphasizes local error signals, making it inherently more suited for real-time, online learning and enabling integration of reinforcement learning and symbolic reasoning. Traditional gradient-based optimization in neural networks struggles to support large-scale real-time learning dynamics due to the brittleness of the underlying backpropagation algorithm, which requires carefully coordinated and synchronized updating across a large network (leading to a reliance on large batch-based updates), and which suffers convergence problems when neural architectures are too recurrent or otherwise too complex. These shortcomings lead to an unfortunate sociotechnical dynamic in which neural architectures oriented toward robust AGI-oriented learning, reasoning, and memory are insufficiently pursued because they tend to involve network topologies for which backpropagation will not easily converge. ActPC resolves these issues on both conceptual and mathematical levels, but can also suffer from lengthy convergence times and undesirable transient dynamics. Information-geometrically enhanced ActPC (ActPC-Geom) provides a compelling potential alternative: by incorporating measure-dependent operators derived from the Wasserstein distance, one aligns parameter updates with the natural structure of the probability distributions underlying ActPC, thereby accelerating learning and smoothing out digressive transient dynamics that might otherwise occur. Like most practical applications of information geometry, ActPC-Geom faces significant computational challenges. However, we believe these can be addressed via appropriate deployment of machine learning and reasoning algorithms within the geometric modeling itself (using neural approximators and information-geometry-guided kernel-PCA-based embeddings). Further, these ML and MR algorithms can be used to inject valuable cognitive properties into the neural learning process, alongside providing acceleration. Read the full paper to learn more: https://t.co/2s4JBsI70F

We are excited to announce that our CEO, Dr. @bengoertzel, COO, @JanetAdamsAI, and the Desdemona robot will be speaking at the Businessabc AI Global Summit (June 26-27) at the Royal Kensington Town Hall in London. Learn more and register to attend: https://t.co/zn5Otr0cAn

Join us this Friday, June 13, at 15:30 UTC, for a new MeTTa Coders session to connect with the Hyperon community and discuss the latest advancements in the MeTTa AGI language. Add our biweekly meetings to your calendar and stay updated on future sessions: https://t.co/77M6JJp65v https://t.co/8va7cYXUcV

We are excited that the @ASI_Alliance's technology stack continues to attract institutional interest and demonstrate the potential for decentralized AI integration across various industries.

As the promise of AGI increasingly captures the world's imagination, we must ensure that the advancement of AI benefits everyone, particularly underserved populations facing persistent educational and economic disparities. iCog Labs, co-founded by Dr. Ben Goertzel and @GETNETASEFFA in 2013 as Ethiopia's first AI company and still by far its most substantial, provides lessons that reveal both the transformative potential and nuanced challenges of applying AI technologies in the developing world. While AI's potential as an educational equalizer is profound, underserved populations often encounter two core challenges: linguistic barriers and culturally irrelevant educational content. UNESCO estimates 40% of students globally lack access to education in a language they understand, yet developed-world tech companies have little motivation to perfect language technology for populations with minimal purchasing power. iCog Labs has pioneered practical solutions. Their collaboration with Curious Learning exemplifies this approach by leveraging generative AI to develop local-language reading apps, which have over 85,000 active users. Additionally, iCog Labs launched Leyu, a decentralized crowdsourcing platform that collects linguistic resources from disconnected communities, gathering data such as parallel spoken sentences that local developers can use to train translation models. Beyond language barriers, effective education demands cultural relevance. Imported educational content frequently fails to resonate with learners whose everyday experiences differ drastically from standardized curricula. The Digitruck project, an off-grid mobile education center deployed by iCog Labs and partially sponsored by SingularityNET, demonstrates this by bringing coding and AI concepts to rural Ethiopian communities through hands-on experience with tablets and maker kits. Students encounter these technologies through applications in relatable contexts, such as improving farming practices, illustrating AI's power to render other technologies practically empowering. These successes highlight a fundamental challenge: current AI development is dominated by a handful of large corporations from two major nations, which explains why AI language technology currently ignores most African languages and serves affluent urban professionals rather than the rural poor in Africa, Central Asia, or elsewhere. The path toward equitable AI-enhanced education requires intentionality, cultural sensitivity, and participatory governance, but the potential rewards of eliminating educational barriers and empowering communities worldwide make this journey imperative. Learn more in @betelhem_dessie, CEO of iCog, and Dr. Ben Goertzel's article: https://t.co/C41h41lMDO

Our CEO, Dr. @bengoertzel, joins @HMsheikh4, CEO of @Fetch_ai, and Nicola Massella, Partner at @STORM_Partners, at Proof of Talk.The fireside discussion explores how the @ASI_Alliance is advancing the development of decentralized AGI. https://t.co/PzdLUE16W8

We are pleased to announce that our CEO, Dr. @bengoertzel, will be speaking at the World Summit AI USA conference at Fort Mason Center in San Francisco (June 18-19). Learn more and register to attend: https://t.co/E45XXMugxh https://t.co/6OG7krGTHi

We are presenting PICO (Prompt Isolation and Cybersecurity Oversight), a new transformer architecture designed to prevent prompt injection attacks and ensure secure, reliable response generation in today's large language models (LLMs). Current defenses against prompt injection—where malicious instructions can override model behavior—often mix trusted system prompts with untrusted user inputs, creating vulnerabilities that are easily bypassed. PICO addresses this through architectural innovation: completely separating system instructions and user inputs into distinct processing channels. The system prompt pathway remains frozen and immutable, while a gated fusion mechanism dynamically weighs inputs using signals from a Security Expert Agent and Cybersecurity Knowledge Graph. We analyze PICO’s effectiveness against sophisticated attacks, including "Policy Puppetry," where malicious instructions are disguised as configuration files. Our mathematical formulation ensures that under adversarial conditions, the trusted system instructions remain dominant in the final output. This work contributes to building safer, more robust AI systems as language models become more widely deployed. We invite the research community to collaborate on the empirical evaluation of the PICO framework and the refinement of adversarial training techniques to validate and enhance the model’s robustness: https://t.co/JK5PjhC1si

CEO Dr. @bengoertzel shares his perspective on AGI and the future of work: “Older people in powerful positions can protect their roles—and they’re the ones controlling how AI is rolled out. So, of course, they’re not going to replace themselves with AI.” https://t.co/1NhxfiXzwb

The Mind Children team will join us from June 18 to 19 at the @WorldSummitAI in San Francisco to showcase their humanoid robot prototype in our booth. Learn more about the event and register to attend: https://t.co/2iL8SmsI20 https://t.co/J46dC929nd

THE TEN RECKONINGS OF AGI are ten critical, unanswered questions that will shape the future of humanity and our relationship with artificial intelligence, from narrow AI to human-level Artificial General Intelligence (AGI) to Superintelligence. This initiative builds on our new global study, which revealed in its first phase mounting public concern over AI governance. Initial findings from US-based respondents indicate that 48% believe advanced AI systems are being developed primarily by and for the benefit of a select few Big Tech corporations and governments, with insufficient public oversight and accountability. The data also shows that two-in-five (39%) Americans are worried that current AI development is not transparent and accountable to the public. The majority (54%) say they feel the direction of development is already out of our hands, and they lack control over the role AI will play in their lives over the next 5-10 years. CEO Dr. @bengoertzel challenges this trajectory, advocating for open dialogue and interoperability among all AI actors to ensure that future developments serve humanity and other sentient beings. Starting next week, each of THE TEN RECKONINGS will be interrogated by Dr. Goertzel together with the world’s leading minds in AI, technology, philosophy, and ethics, with publicly available information for all to engage with the discussion. Follow THE TEN RECKONINGS on YouTube and subscribe to stay updated as each question is explored: https://t.co/CYVGMBZCNe

Join us at the world's oldest and most prestigious gathering dedicated exclusively to general machine intelligence research: The 18th Annual Conference on Artificial General Intelligence (AGI-25). AGI-25 will convene a worldwide community of researchers and developers, including notable figures like Ben Goertzel, Richard Sutton, Tatiana Shavrina, Henry Minsky, and Kristinn R. Thórisson, all working on the latest innovations toward generally intelligent machines—the next evolution of AI. This year’s program will include mainstage keynotes and technical talks, hands-on workshops and tutorials, advanced software and hardware demonstrations, networking opportunities within our global community of innovators, and immersive experiences. Register now to attend in person or watch online: https://t.co/GRxMnwHk7Z

Simply asking narrow AI systems to 'work super hard' on AGI problems reveals a fundamental misconception about the nature of AGI research. The deep technical challenges involved in building AGI, such as optimizing Hyperon's MeTTa Optimal Reduction Kernel (MORK) for scaling neural-symbolic-evolutionary AI, require sophisticated problem-solving and deep reasoning that LLMs do not possess. This does not imply, nonetheless, that there is nothing to learn about human-level AGI from studying and experimenting with LLMs, nor that LLMs cannot form significant parts of human-level AGI architectures that also incorporate other ideas. The path to achieving true AGI, how to bring LLMs slightly closer to human-like cognition, and other important questions will be discussed at the 18th Annual AGI Conference from August 10-13: https://t.co/c4QT7N1I0v

CEO Dr. @bengoertzel in 2020: "The idea that you can take methods that have worked for particular narrow AI problems and scale them up to achieve AGI by adding more processor power and more data — just isn’t gonna work." The path forward is OpenCog Hyperon.

The Businessabc AI Global Summit – London, England (June 26-27)This exclusive premier international summit will unite over 50 global keynote and panel speakers and an international network of 100+ universities for a forward-thinking dialogue around AI: https://t.co/lw1AGOTJDv

We maintain a strong presence at major industry events worldwide, advancing discussions on decentralized AI and the development of beneficial AGI. Throughout June 2025, our team will be participating in strategic discussions at the following events ↓ https://t.co/Wu4GC2H9Jc

We are pleased to announce that our Chief AI Alchemist, Dr. @Mihaela_Ulieru, will speak on "AI for Impact: from Promise to Practice" at the Accelerate 2050 Conference in Brindisi, Italy, on June 10. Learn more and register to attend: https://t.co/8JIhOTxm2i

Homomorphic encryption (HE) delivers cloud-scale compute power with rigorous end-to-end data privacy. While classical HE schemes support encrypted search, analytics, and limited machine learning workloads, two critical gaps remain: - Post-quantum security challenge: Most mature HE systems rest on RSA, elliptic-curve, or pairing assumptions–all vulnerable to Shor-style attacks once large, fault-tolerant quantum computers arrive. - Quantum computing limitation: None of the widely deployed schemes allow a user to outsource quantum computations (e.g., variational kernels, error-correction gadgets) while keeping both the algorithm and the quantum data hidden. Our research addresses both gaps simultaneously, building on the categorical HE framework for intuitionistic proofs and total functional programs introduced in Dr. @bengoertzel's paper on "Homomorphic encryption of intuitionistic logic proofs and functional programs: A categorical approach inspired by composite-order bilinear groups" and extending it to the quantum domain: https://t.co/l5lZdtMHt7

Homomorphic encryption (HE) delivers cloud-scale compute power with rigorous end-to-end data privacy. While classical HE schemes support encrypted search, analytics, and limited machine learning workloads, two critical gaps remain: - Post-quantum security challenge: Most mature HE systems rest on RSA, elliptic-curve, or pairing assumptions–all vulnerable to Shor-style attacks once large, fault-tolerant quantum computers arrive. - Quantum computing limitation: None of the widely deployed schemes allow a user to outsource quantum computations (e.g., variational kernels, error-correction gadgets) while keeping both the algorithm and the quantum data hidden. Our research addresses both gaps simultaneously, building on the categorical HE framework for intuitionistic proofs and total functional programs introduced in Dr. @bengoertzel's paper on "Homomorphic encryption of intuitionistic logic proofs and functional programs: A categorical approach inspired by composite-order bilinear groups" and extending it to the quantum domain: https://t.co/l5lZdtMHt7

The @ASI_Alliance is hosting an investor-focused event on decentralized AGI development in Paris on June 11th (5:30-7:30 PM CEST). Speakers include: - Dr. Ben Goertzel - CEO, ASI Alliance & SingularityNET - Janet Adams - Board Director, ASI Alliance & COO, SingularityNET - Mario Casiraghi - Executive, ASI Alliance & CFO, SingularityNET - Humayun Sheikh - Chairman, ASI Alliance & CEO, Fetch․ai - Matt Hawkins - Member, ASI Alliance & CEO, CUDOS The team will share updates on strategy, key initiatives, and the technical roadmap. The evening features a technical presentation, followed by a Q&A session and exclusive networking opportunities with visionary leaders and members of the ASI Alliance ecosystem. Registration is limited for investors and institutions interested in AI investment opportunities and is subject to review: https://t.co/i7FxUHaI9c

OpenCog Hyperon is the most advanced open, decentralized framework for artificial general intelligence (AGI) at the human level and beyond.Read CEO Dr. @bengoertzel's recent interview with Cointelegraph: https://t.co/05koVqIYzq https://t.co/dYWiQsJHon

The AIRIS (Autonomous Intelligent Reinforcement Inferred Symbolism) agent we deployed within the Minecraft 3D gaming environment establishes one of the first intelligent systems capable of autonomous, adaptive learning.Get started now: https://t.co/0o6ArA8Q0J https://t.co/cMKk3GSulY

Join us this Friday, May 30, at 15:30 UTC, for a new MeTTa Coders session to connect with the Hyperon community and discuss the latest advancements in the MeTTa AGI language. Add our biweekly meetings to your calendar and stay updated on future sessions: https://t.co/77M6JJoyfX https://t.co/yxtnryXxFk

CEO Dr. @bengoertzel sat down with Gus Docker on the Future of Life Institute podcast to discuss: - What distinguishes the current AI boom from previous ones - Important but overlooked AI research - Simplicity versus complexity in the first AGI - The feasibility of alignment - AGI benchmarks and economic impact - Potential bottlenecks to Superintelligence - What humanity should do moving forward. Watch here: https://t.co/qQyE7H9pe6

The application deadline for our $830,000.00 AGI R&D grants is approaching. This is your opportunity to participate in shaping the course of AGI and Superintelligence. Review the RFP requirements and submit your proposal today at https://t.co/h6yjXlMiTl https://t.co/3vWzGp7jPT

"SingularityNET and the @ASI_Alliance are “probably the only serious AGI R&D team outside of Big Tech, certainly the only one in the crypto space" — Via @Cointelegraph https://t.co/05koVqIYzq

CEO Dr. @bengoertzel's AI Twin, powered by @twinprotocolAI, invites you to apply for our $830,000 AGI R&D grants. Learn more and submit your proposal by May 27, 2025, at 23:59 UTC https://t.co/h6yjXlMiTl https://t.co/zepioBftvk

The AI Twin of our CEO, Dr. @bengoertzel, powered by @twinprotocolAI, invites you to apply for our $830,000 AGI R&D grants. Learn more and submit your proposal by May 27, 2025, at 23:59 UTC, via https://t.co/h6yjXlMiTl https://t.co/mk2vyCUMKj

Watch our COO, @JanetAdamsAI, discuss the combined values of AI and blockchain, OpenCog Hyperon, our compute infrastructure, and how decentralized AI creates new possibilities for humanity: https://t.co/54PYObEb52

In the MeTTa AGI programming language, functions work through pattern matching against equality expressions. When you call a function like (only-a A), it's essentially executing a query that matches against equality definitions. A key difference in MeTTa's approach is how unmatched expressions are handled: while direct match operations produce empty results when no match is found, the interpreter keeps unmatched expressions unreduced, performing additional processing on equality queries. MeTTa's unification function allows two patterns with variables to be unified, matching them together so shared variables receive consistent bindings. This function evaluates to its success argument when unification succeeds and its fallback argument when it fails. Using match within functions creates convenient functional abstractions while preserving the declarative nature of knowledge. As the documentation shows with "get-parent-entries" and "get-parents" examples, this approach is useful for representing and processing knowledge in the program Space. Visit https://t.co/Ys1fJs2IJm to learn more about MeTTa's pattern matching and unification capabilities.

Update: The application deadline for our AGI R&D grants has been extended by 7 days. The revised submission period now concludes on May 27, 2025, at 23:59 UTC, providing additional time for academic researchers, independent developers, and specialized technical teams to finalize their proposals. Learn more and submit your project via the Deep Funding portal at https://t.co/zZ2CVR2GL1

We've been working on a solution to this exact problem. More to come very soon. @ASI_Alliance

The application deadline for our AGI R&D grants is rapidly approaching. This is your opportunity to participate in shaping the course of AGI and Superintelligence. Our grants support projects across six critical domains: concept blending in MeTTa, advanced knowledge graph tooling, neural-symbolic integration, novel hardware architectures, theoretical quantum computing models, and AGI motivation systems. Learn more and submit your proposal today via https://t.co/zZ2CVR2GL1

CEO Dr. @bengoertzel joined @Rachelwolf00 on the Web3 Deep Dive podcast to discuss the path from narrow AI to human-level AGI, the importance of decentralization, our recently launched "10 Reckonings of AGI" initiative, longevity, digital twins, and more: https://t.co/unrr4jVG59 https://t.co/oCRPjJIPxH

Just two days remain to apply for our $830K AGI R&D grants advancing the OpenCog Hyperon framework. These grants address critical technical challenges across six essential domains for beneficial AGI development, providing a unique opportunity for researchers and developers to contribute to open, decentralized artificial general intelligence. Academic institutions, independent researchers, and specialized teams are all invited to submit proposals via https://t.co/zZ2CVR2GL1

Just two days remain to apply for our $860K AGI R&D grants advancing the OpenCog Hyperon framework. These grants address critical technical challenges across six essential domains for beneficial AGI development, providing a unique opportunity for researchers and developers to contribute to open, decentralized artificial general intelligence. Academic institutions, independent researchers, and specialized teams are all invited to submit proposals via https://t.co/zZ2CVR2GL1

As Big Tech companies continue channeling billions of research dollars and attention into developing and scaling LLMs, a fundamental question is emerging in tech circles: Are LLMs actually leading us away from true AGI? At a #Consensus2025 panel addressing whether Web3 is losing the AI race, our CEO, Dr. @bengoertzel, challenged the conventional wisdom driving investment strategies at many leading AI companies and sovereign wealth funds, including the recent multi-billion-dollar investments from Saudi Arabia and the UAE in US AI infrastructure. "I would quote Yan LeCun, a pioneer of deep learning and the head of AI at Facebook, who said on the highway to AGI, LLMs are an off-ramp," Dr. Goertzel told attendees, rejecting the premise that Web3 approaches are falling behind centralized AI development: "If you've gotten off the off-ramp, it doesn't matter if you're going 1,000 miles an hour and the other guy's only going 300 miles an hour if they're going on the right highway to the destination." The panel, which featured Ben Fielding (Founder, Gensyn), Jesus Rodriguez (CEO, IntoTheBlock), Clara Tsao (Founding Officer, Filecoin Foundation), and Jeff Wilser (Founder and Host, The People's AI Podcast), revealed a stark divide in how industry experts view the future of decentralized AI. While other panelists pointed to Web3's current disadvantages in talent, datasets, and infrastructure, Dr. Goertzel addressed a deeper issue: the incremental improvement of LLMs is not a viable approach to achieving human-level AGI. However, "If scaling up transformer neural nets is the crux of how you get to AGI, it's hard to see how the US and Chinese governments and the Big Tech companies in their orbit don't win the race," Dr. Goertzel acknowledged, noting the immense capital these entities are deploying. Our AGI R&D efforts at SingularityNET suggest a different path forward. "My own research intuition is that LLMs are not suited to be the central hub of a human-level AGI, let alone a Superintelligence, although they can be a powerful ingredient in a hybrid architecture for AGI," Dr. Goertzel explained. Our team is developing OpenCog Hyperon, a "hybrid, deep neural net, symbolic reasoning, evolutionary learning, approach to AGI within more sophisticated cognitive architectures than LLMs comprise," as described by Dr. Goertzel. This cognition-level approach represents a fundamentally different direction from most mainstream AI development. Dr. Goertzel closed with a prediction that would have seemed outlandish just a few years ago but now reflects our growing confidence in decentralized approaches through which AGI will be in the hands of humanity at large and without a single owner or controller: "Within the @ASI_Alliance, we will launch the first AGI within one to three years from now on a decentralized infrastructure, and Big Tech will play catchup."

There are just 3 days left to apply for our $830K AGI R&D grants dedicated to advancing the OpenCog Hyperon framework. Review the RFP requirements and submit your propsoal today at https://t.co/h6yjXlMiTl https://t.co/8J7hpVP2rW