Jonathan Siddharth(@Jonsid):To get to ASI we likely need auto-meta-research, not just auto-research. Auto-research hill climbs within the current recipe. Minimize pretraining loss, maximize post-training evals. Auto-meta-research defines new objectives. An outer loop that searches across paradigms. Outside deep learning, maybe even outside gradient descent. Not just scaling transformers + RL. The inner loop optimizes the recipe. The outer loop questions the recipe.

Jonathan Siddharth

@Jonsid

Founder and CEO, Turing. Accelerating superintelligence to drive real economic progress.

加入 November 2008

4.9K 正在關注 8.1K 粉絲

Jonathan Siddharth@Jonsid

2026.09.13 20:19

To get to ASI we likely need auto-meta-research, not just auto-research. Auto-research hill climbs within the current recipe. Minimize pretraining loss, maximize post-training evals. Auto-meta-research defines new objectives. An outer loop that searches across paradigms. Outside deep learning, maybe even outside gradient descent. Not just scaling transformers + RL. The inner loop optimizes the recipe. The outer loop questions the recipe.