The Man Behind AlphaGo Thinks AI Is Taking the Wrong Path

Editor's take

Demis Hassabis's former colleague, David Silver, has launched a new venture, radicalizing his approach to artificial intelligence development. Silver’s company, reportedly backed by $1 billion in funding, is pursuing a vision of "superlearners" that diverges from the path currently dominated by large language models like OpenAI's GPT-4.

This shift is significant because it challenges the prevailing paradigm of generative AI, suggesting that scaling current architectures may not be the most effective route to advanced AI capabilities. Silver’s expertise, honed during his pivotal role in AlphaGo's development, implies a focus on reinforcement learning and potentially novel architectures that prioritize deeper understanding and more efficient learning over sheer parameter count. This could impact research directions and investment in the broader AI landscape, potentially creating a new frontier beyond LLMs.

Future developments to monitor include the specific architectural innovations Silver’s team will unveil and their demonstrable performance against established benchmarks. The success of this "superlearner" approach will hinge on its ability to achieve breakthroughs in areas where LLMs currently struggle, such as robust reasoning, long-term planning, and true generalization, potentially setting a new benchmark for AI progress.