AI errs 1-5% — but humans do too, argues dev: apply the same checks to both
aronchick · x · 2026-10-05
aronchick pushes back on atomkirk's question — "AI does stupid stuff 1-5% of the time; what low-stakes work tolerates that?" — arguing the premise is flawed: humans themselves make mistakes 5% of the time and are routinely corrected by machines and systems. His takeaway: if we apply the same oversight and correction tools to AI that we already apply to humans, we'll be fine.
A notable viewpoint clash on AI error tolerance: one side sees a low error rate as disqualifying AI from high-stakes work, the other argues human error rates are similar and systematic correction mechanisms are the real answer.
More from AGI Musings
- AI safety's disconnect from systems safety engineering questioned; FMF cites early INPO ties — chrismeserole · 2026-10-05
- LLMs are pure map with no territory: why you can't make a model of a model — inductionheads · 2026-10-05
- LLM Progress Called Incremental, Not Exponential, as Narrative Debate Heats Up — inductionheads · 2026-10-05
- Gary Marcus backs WSJ skepticism: 9%-of-GDP AI spending faces a natural ceiling — GaryMarcus · 2026-10-05
- Timnit Gebru wins 'Alternative Nobel', calls AI existential-risk talk a ploy to dodge regulation — PolarBearby · 2026-10-05
- Would COVID vaccines ship faster in 2026 with OpenAI and Anthropic in the loop? — flowersslop · 2026-10-05