Ben Goertzel, founder of SingularityNET, has been working on artificial general intelligence (AGI) for over two decades. His current project, OpenCog Hyperon, is a unique AGI system designed to function as an autonomous, self-aware agent. Goertzel claims the alpha version of this system is already demonstrating a degree of self-awareness, setting it apart from other AI models like chatbots.
Key Takeaways:
- OpenCog Hyperon: Autonomous AGI system, demonstrating early self-awareness.
- Development Timeline: In progress since 2001, Alpha version launched in April.
- Scalability: Major rebuild for improved speed and performance by end of the year.
- Decentralized AI: Merging with CUDOS for increased computational resources.
- Future Vision: Combining centralized and decentralized infrastructure for global AGI development.
OpenCog Hyperon has been in development since 2001, with its first open-source launch in 2008. The system is different from traditional large language models (LLMs), such as GPT-4, as it integrates logical reasoning, evolutionary learning, and dynamic knowledge graphs. This approach allows the system to adapt and update itself, paving the way for true AGI.
One of the key goals of the project is scalability. Goertzel’s team started a complete rebuild of OpenCog three years ago to achieve massive scalability. The alpha version was launched in April, but Goertzel notes that the system is still slow and in need of optimization.
The goal is to significantly speed it up, with improvements expected by the end of the year. Once this phase is complete, the team will focus on building toward a fully functional AGI on the Hyperon infrastructure.
The Beginning of The Artificial Superintelligence Alliance (ASI)
The Artificial Superintelligence Alliance (ASI) was formed in March, bringing together projects like SingularityNET, Ocean Protocol, and FetchAI. This alliance supports the development of AGI systems, especially in decentralized and open-source environments. A recent merger with the CUDOS decentralized cloud hardware network was approved by 96% of voters, giving Goertzel more computational power to further enhance OpenCog Hyperon.
OpenCog Hyperon follows a different path from models like OpenAI’s latest o1 model. According to Goertzel, OpenAI’s decision not to turn its new model into an autonomous agent may have been driven by regulatory concerns. He believes regulators could view autonomous AGI as too risky and potentially impose strict regulations. In contrast, Goertzel’s decentralized, open-source approach makes it harder for any single country to control or ban AGI development.
Advantages of Decentralized AI Systems
One of the main advantages of decentralized AI systems is that they operate across many machines and locations globally. This reduces the risk of centralized control and opens up the possibility for broader development. However, decentralized AI faces its own challenges, particularly in terms of cost and energy requirements. Training large neural networks is expensive and energy-intensive, which can be a barrier to scaling decentralized systems.
To address these challenges, Goertzel’s team plans to combine centralized and decentralized infrastructure. They have allocated a significant portion of their budget to acquiring powerful GPUs to support their work. The decentralized aspect will gradually expand, drawing in additional computing power from connected machines through platforms like NuNet and Hypercycle.
Goertzel’s vision for OpenCog Hyperon is to make AGI accessible and useful to everyone. By creating a decentralized system, he hopes to avoid the pitfalls of centralization and build a robust AGI that can contribute to a future shaped by intelligent, autonomous systems.
Disclaimer
FAQ
DeFI stands for decentralized finance, offering open and accessible financial systems built on blockchain technology.
Yield farming involves earning interest by lending or staking cryptocurrencies.
Layer 1 blockchains are the primary networks (e.g., Ethereum), while layer 2 blockchains scale and improve performance on top of them.