AI safety researcher flags shaky 'time of perils' assumption in 'more AI is the fix' arguments
davidmanheim · x · 2026-11-01
Responding to a call to lay out the cruxes of "the only way out is through" arguments — the idea that current AIs are already dangerous enough (in cyber and bio) that only more AI development can secure things — David Manheim says the reasoning relies on a shaky "time of perils" assumption. He notes MacAskill outlined one problem but understates the issue: the assumption also presumes eventual defense dominance, which is unverified.
Related event: AI safety researchers debate 'the only way out is through' argument(2 posts)→
More from AGI Musings
- Charting the US-Europe GDP gap if Amodei's 10–20% AI growth premise holds — ChrisGPT · 2026-11-01
- Prediction: typing will feel like using fax machines in five years — theteknosaur · 2026-11-01
- Redditor warns: giving AI a survival drive is 'the most obvious wrong path' — Lonelyfurry818 · 2026-11-01
- 10 spicy takes from 50 frontier robotics insiders: Astra is robotics' GPT-3 moment — ZhitingHu · 2026-11-01
- "Sad": LLMs have collapsed the distance between intelligence and rent seeking to zero — jwt0625 · 2026-11-01
- Microsoft offers voluntary exit to 60,000 employees as tech workers' AI anxiety deepens — jackedAJ · 2026-11-01