Google says it has no solution for AI mistakes

futurism.com

5 pointsalangibson4 comments

Seems like there's a fairly straightforward fix for this: hold companies legally liable for the content they they produce and publish using LLMs.

When the companies start getting held responsible for the dangerous or libelous misinformation that they're pushing out, then they'll either find a solution to it or stop doing it. If they're allowed to continue doing so then why would they care?

Since it will never be economical to hire a human legal team to vet every generative output, that is equivalent to banning generative AI from general use.

Sounds good.

It doesn't ban the use of LLM - it just shifts the costs of their errors back to the company rather than them being bourne by the public.

And if companies decide that they're LLMs are only viable if they get to ignore the negative externalities then a reduction in their use sounds like a good thing.

Reply on news.ycombinator.com