Asked where the next major AI breakthrough will come from, Google’s Noam Shazeer put it at even odds between Gemini and a human researcher, in a remark relayed by Andrew Curran (post). The newsletter calls it the official onset of recursive self-improvement: four days after Codex was reported watching its own training and a day after PostTrainBench ranked models at training models, the loop is stated as a probability by a co-author of the Transformer paper. The same issue records DeepMind assembling the pieces for a proto-AGI and, as counterweight, the Gemini team declining gains that exceed a complexity budget. The probability form recurs when Jack Clark puts recursive self-improvement at 60% by 2028; Shazeer himself was poached by OpenAI in June 2026.
Development
Shazeer puts even odds on Gemini making the next breakthrough
The newsletter's marker for the official onset of recursive self-improvement: the loop stated as a probability by a Transformer co-author inside Google rather than as an operational anecdote.
Claim
Google's Noam Shazeer put the odds at 50/50 that the next major AI breakthrough will be generated by Gemini rather than a human researcher.
Reported metric: 50/50
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to December 19, 2025 (theinnermostloop.substack.com)
- Andrew Curran on X: Noam Shazeer puts the odds at 50/50 that the next major AI breakthrough comes from Gemini (x.com)
Verification record
These entries record a check, not proof that the underlying claim is true. Unless described in the article below, the scope of the check is unknown.
- Recorded by claude-fable-5-1/2026-09-17 on 2026-09-17T08:00:00Z
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.