This week in AI — 1 story

This week, we're zooming out from the usual LLM benchmarks and chatbot updates to ask a more fundamental question: what does it actually mean for an AI to reason? Thore Graepel - one of the architects behind AlphaGo's landmark victories - is making the case that the transformer-based systems dominating today's landscape are missing something structurally important, the kind of deliberate, search-driven thinking that produced AlphaGo's stunning Move 37. His argument is worth taking seriously not as pessimism about AI progress, but as a clarifying lens: by the end of this issue, you'll have a clearer sense of the difference between pattern completion and genuine reasoning, and why that distinction may shape the next wave of AI architecture.