Skip to content

AI models don't kill people – people kill people

theregister.com
3 pointsjoebuckwilliams1 comment
On HN

Comments

As ever, identifying danger exclusively with human willpower and detaching it from means at their disposal commits you to absurdities where butter knives, guns, and AGI are indistinguishable risks.

So, I agree entirely with the article, though perhaps my model would be more like social media addiction lawsuits than airbags, though each comparison has their merits. The really interesting implication of this article is that training that's currently happening may be putting Sam Saltman et al on the hook for steep fines and penalties, years or perhaps decades from now as legal paradigms and research catch up to the societal impacts of these models.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.