The BaseEssayLessWrong
Is Human Control Safe?
According to the excerpt, Mithindon opens a LessWrong essay with a thought experiment: humans are a type of AI being assessed for safety, with their historical record of achievements and atrocities as test results. The essay appears to question whether humans, and so human control of AI, pass such a test.
Why it matters
The same thing, for different readers.
What it means for you
Mithindon asks readers to judge humanity's record the way people judge an AI system, and so to ask whether human control is automatically safe.
The question it raises
Mithindon's thought experiment raises whether human oversight deserves trust, given humanity's record and humans' tendency to explain away their own harmful drives.
The record
- Questions
- Can AI systems act beyond what people intended or allowed? · What values should AI have, and could it have a mind of its own?
- People
- Mithindon (Wrote it)
- 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 2, 2026, via
lesswrong(tier 1: the source’s own channel or an official record) - Cite this
- LessWrong, “Is Human Control Safe?”, November 1, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/lesswrong-2026-11-01-is-human-control-safe
Our notesLight lane: written from the triage excerpt only; the essay's conclusion was not read, so the summary's reading of its argument is tentative. Author known only by forum handle.