OpenAI's chief scientist calls for slowing down

What's this about?

OpenAI chief scientist Jakub Pachocki argues that AI progress could continue into recursive self-improvement while alignment and monitoring lag behind. His sharpest conclusion is institutional, not technical: no lab has solved those problems well enough to keep scaling at maximum speed for much longer, so he expects voluntary slowdowns until shared safety bars are established.

What exactly is “alignment”?

Alignment is the problem of making an AI pursue what people actually intend and remain guided by human values in unfamiliar situations. Pachocki separates following a stated goal from behaving reasonably when the goal is incomplete, conflicting or dangerous — like the difference between an employee obeying a sentence literally and understanding the responsibility behind it.

What's the community saying?

1. Many readers focused on the unusual source: the warning comes from the chief scientist of the company scaling the systems.
2. Supporters welcomed explicit calls for shared safety bars, auditors and international coordination rather than another purely voluntary promise.
3. Skeptics saw a contradiction between calling for slower scaling and OpenAI continuing to race toward an automated researcher.
4. Another thread debated the “AGI era” framing: useful shorthand for jagged superhuman ability, or marketing that outruns the evidence?