AI BriefingDavid Silver Emphasizes RL-Based 'Superlearners'
Key point
AlphaGo developer David Silver points out the limitations of LLMs and presents reinforcement-learning-based 'superlearners' as the next generation of AI.
Details
David Silver, the key developer of AlphaGo, argues that current LLMs (ChatGPT, Gemini, etc.) may face limits to their growth because they depend on the limited resource of human-generated data.
Instead, he presents Reinforcement Learning systems that learn through trial and error in simulated environments—so-called 'superlearners'—as the next generation of AI models. Unlike the existing approach of learning from internet data, this method involves discovering new knowledge on its own through repeated experimentation and failure, just as AlphaGo built up its skill through playing games.
To realize this technical direction, Silver has founded a new startup and secured roughly $1.1 billion (about 1.5 trillion won) in investment.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Stanford and NVIDIA Release 'Contrastive Language Model' CLM with 9x Faster Inference