General Discussion

the last few years, and while some were harmless and funny

Cartoonist/author Martin Rowson of the Guardian asked Google's AI to name his wife. The results were hilarious.
https://www.democraticunderground.com/100221008588

some were potentially very dangerous:

'Unbelievably dangerous': experts sound alarm after ChatGPT Health fails to recognise medical emergencies (The Guardian)
https://www.democraticunderground.com/100221066192

The maddening thing about LLM hallucinations is that the bots can sound so convincing that people tend to trust them, and verifying accuracy and proofreading is time-consuming, and people often are using AI primarily to save time. I understand being tempted to save time.

But I don't know if we'll ever be able to clean up all the AI slop that's already flooded the internet in just a few short years. And the stories I've read about security risks from AI-generated code have me hoping some large company will have enough trouble from AI code to scare most people away from using it, before there are really catastrophic hacks and failures.