The Rise of Self-Aware AI: Insights from Ben Goertzel
Introduction to OpenCog Hyperon
Ben Goertzel, the founder of the Artificial Superintelligence (ASI) Alliance, has revealed that the alpha version of OpenCog Hyperon, an artificial general intelligence (AGI) system, has attained a form of self-awareness. This marks a significant milestone in AI development, as the system is not merely a chatbot but an autonomous agent with self-awareness.
Autonomous AI: The Vision and Reality
Goertzel believes that OpenAI, despite its advancements, has hesitated to fully embrace autonomous agents due to potential regulatory backlash. OpenCog Hyperon, however, is designed to be an independent entity with its own goals and awareness. This difference sets it apart from other AI models like GPT-4.
The Formation of ASI Alliance
In March of this year, the ASI Alliance was formed, bringing together Goertzel’s SingularityNET project, Ocean Protocol, and FetchAI. The merger with CUDOS, approved by 96% of voters, will enhance the computing power available for scaling up OpenCog Hyperon.
The Future of AGI with OpenCog Hyperon
Three years ago, OpenCog embarked on a comprehensive rebuild aimed at achieving massive scalability. The alpha version launched in April, and despite initial slow performance, the team is working on significant speed improvements. Goertzel anticipates substantial progress by next year, enabling the pursuit of AGI on the new Hyperon infrastructure.
Self-Awareness in AGI Systems
Goertzel explains that the current model of OpenCog Hyperon possesses a form of self-awareness. Unlike traditional chatbots, it has a model of itself and its interaction with users, striving to achieve specific goals. This approach integrates logical reasoning, evolutionary program learning, and deep neural networks within a dynamic knowledge graph.
The Challenge of Autonomous AI
Goertzel notes that the lack of a world model has hindered the success of autonomous AI agents. Simply transforming a question-answering system into an interactive agent is insufficient. OpenCog Hyperon addresses this by being an autonomous agent from the outset.
Regulatory Risks and OpenAI’s Approach
Goertzel believes OpenAI has deliberately avoided developing fully autonomous agents to mitigate regulatory risks. OpenCog Hyperon, on the other hand, is being developed as a decentralized, open-source system, making it resilient to regulatory crackdowns.
The Decentralized Approach
Developing decentralized AI systems presents challenges, particularly in terms of compute power and cost. However, Goertzel is optimistic about the feasibility of running logical reasoning and evolutionary learning algorithms on a decentralized network. The ASI Alliance is investing in GPUs to create supercomputers, and additional computing power can be sourced from SingularityNET spin-offs.
Renewable Energy for Decentralized AI
The significant power demands of AI systems have led the ASI Alliance to explore renewable energy sources. Discussions are underway with governments in Ethiopia and Paraguay to establish AI data centers near dams, providing substantial power at lower costs.
Conclusion
The development of OpenCog Hyperon represents a significant step towards achieving AGI. By integrating self-awareness and autonomy, and leveraging decentralized networks and renewable energy, the ASI Alliance aims to create a smarter and more resilient AI system. As these advancements unfold, the potential for AGI to transform various fields continues to grow, heralding a new era in artificial intelligence.
