Sci-fi idea: Safety-pilled algorithms researcher vs hardware engineer on AI risk

kellerjordan0 · x · 2026-08-11

Author @kellerjordan0 shares sci-fi idea: a safety-pilled algorithms researcher argues for rapid algorithmic progress to learn if current compute is dangerous; a safety-pilled hardware engineer argues for rapid compute buildout to learn if current algorithms are dangerous. The author suggests the answer depends on which progress is harder to freeze, likely the algorithms researcher.

Related event: The Algorithm vs. Hardware Paradox in AI Safety(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →