The BaseEssayLessWrong
Can alignment realistically be solved under Plan C?
According to the excerpt, Simon Lermen responds to Scott Alexander's argument that plans giving a few months to solving alignment deserve serious consideration. Lermen argues that a short pause cannot solve superintelligence alignment, because there is one critical shot and little room to learn from experience.
Why it matters
The same thing, for different readers.
For lawmakers
Simon Lermen's argument bears on how long any pause would need to be, and on whether delegating alignment to AI systems within months is a credible plan.
The question it raises
The post disputes who carries the burden of proof on whether alignment is hard, and when deferring to experts counts as reasoning.
The record
- Questions
- Can AI systems act beyond what people intended or allowed? · How fast is AI advancing, and where is it heading?
- People
- Simon Lermen (Wrote it) · Scott Alexander (Mentioned)
- What we read
- We have read only the description the publisher gives. No quotations until someone reads the source itself.
- Found by
- The Radar, run of November 4, 2026, via
lesswrong(tier 1: the source’s own channel or an official record) - Cite this
- LessWrong, “Can alignment realistically be solved under Plan C?”, November 3, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/lesswrong-2026-11-03-alignment-under-plan-c
Our notesLight lane: written from the triage excerpt only; no source text fetched, no quotations. Scott Alexander's post is described only as the excerpt describes it.