TBPN Metis List expands to 128 AI researchers: Noam Shazeer takes the top spot, Yann LeCun drops off

Summary

TBPN expanded its Metis List of influential AI researchers from 100 to 128 names. The update reshuffles the top tier and reveals where AI research power now concentrates.

Top Five Reshuffle

Noam Shazeer moves to number one, displacing Ilya Sutskever. Sutskever has published little at OpenAI recently, while Shazeer has sustained strong output. Gemini's quality improvement tracks directly to Shazeer's return from Character AI. The new top five are Shazeer, Sutskever at two, Demis Hassabis at three, Dario Amodei at four, and John Schulman at five. Hassabis was missing from the original list, a decision the curators acknowledge may have been a mistake. He still shapes research direction at DeepMind even as his CEO role demands his time. Everyone in the top five except Hassabis has worked at OpenAI at some point.

Major Movers

Noam Brown jumped 36 spots to number 13, credited for post-training and reinforcement learning work at OpenAI. Peter Beale, a Berkeley academic, climbed 66 spots and influences research despite being outside a major lab. Mark Chen broke into the top 10 at number six, up 19 spots.

Yann LeCun fell entirely off the list after previously ranking in the top 10. He remains Meta's chief AI scientist formally, but his influence has visibly declined as Mark Zuckerberg built out the Meta Super Intelligence Lab around other leaders. LeCun's skepticism toward LLMs and loss of institutional power at Meta appear to have moved the needle. Jürgen Schmidhuber dropped 53 spots, partly due to his controversial and failed attempt to claim credit for work that led to the Nobel Prize.

Lab Distribution

OpenAI, DeepMind, and Anthropic are now neck-and-neck at the top of researcher count. OpenAI has 24 researchers on the list, DeepMind 23, and Anthropic 22. In the previous list, Anthropic led by roughly four researchers. Thinking Machines, Mira Murati's company, has 12 researchers on the list, a notable performance for a young firm. Meta ranks at eight, still in early stages of building out the MSL despite recent hiring blitzes. The sparse number reflects that the lab only launched months ago. If the list had been made when MSL started, Meta would have had nearly zero apart from LeCun.

Geography

The list skews heavily American, with 51 researchers from the USA. The UK follows with 13, Canada with 12. By population-weighted density, Canada punches well above its weight, roughly one-tenth America's population but one-fifth as many top-tier AI researchers on the list. China remains opaque. The curators identified two DeepSeek researchers and flagged that many Chinese labs operate in secrecy, making comprehensive counting difficult. No researchers from ByteDance appear despite the company's prominence in recommendation algorithms.

Methodology Shift

The curators acknowledged over-optimizing the previous list for Twitter cloud and Google Scholar citations. Modern labs are secretive and rarely publish. Researchers face active incentives not to do press because high visibility invites poaching. Some teams even requested removal from the list to reduce external attention. This forced a rebalance toward insider knowledge and direct impact assessment rather than paper count alone.

Caveats

The list is still live and subject to change. Critics have already flagged omissions. Rich Sutton and John Carmack, both at Keen, were noted as probable inclusions. George Boole and Alan Turing were cut entirely, a decision that drew pushback. The curators preserved one Easter egg with Leibniz on the list but avoided going full meme. The exercise remains contentious because it directly affects poaching incentives and researcher visibility in a hypercompetitive talent market.