Hacker News new | past | comments | ask | show | jobs | submit login

So organizations won't deploy AIs to provision resources?

Executive: "We can save money by using AI to provision these resources more efficiently."

Executive 2: "Yeah, but it would be dangerous to give an AI control over these resources."

Executive: "Oh, right, forget I said anything. What's the next topic on the agenda?".




Not for decision making, and I've had these conversations with peers.

Liability is still very much a thing, and sentence extruders hallucinate, leaving you open to liability because you still need to show a decision chain.

This is why lawyers are getting disbarred (edit: formally sanctioned by Bar Associations. Repeated violation becomes cause for suspension or disbarment) for using LLMs and plenty of PMs and SWEs getting pink slipped for unauthorized LLM use.


Which lawyer have been disbarred for using LLM?



Even so, they would need access to money and the power grid to truly scale to a dangerous degree.


You're fearmongering. The single most-dangerous credible threat such an AI would pose is spending $400,000 on EC2 Postgres instances.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: