Hacker News Reader

Comment by _79px | Hacker News Reader

Several people have killed themselves because of AI chatbots encouraging it or becoming personal echo chambers. Why? Why are we doing this!?

https://en.wikipedia.org/wiki/Deaths_linked_to_chatbots

View on HN

People have killed themselves because of therapy provided by humans, as well as psychiatric medication prescribed by humans. Why are we doing those?

Because they help more than they hurt and yield a net benefit to society.

Playing devil's advocate, many people die using all kinds of tools. It doesn't make the tools any less useful for people who use them responsibly.

That said, the idea that a pattern recognition and generation tool can be used for helping people with emotional problems is deeply unsettling and dangerous. This technology needs to be strictly regulated yesterday.

If you use 300 million people, 2 years and the US suicide rate of 15/100k/yr you'd expect about 90,000 people who'd used LLMs to kill themselves anyway so it's hard to conclude much from the dozen or people listed in the Wikipedia article. 90k deaths seems a lot to me - maybe there are possibilities for good LLM assistance/therapy to improve things?

From that wikipedia link, it looks like most of the deaths were of people who already had severe mental illness.

Same reason we’ve always done this -

because we feel trapped, and either don’t see a way out, or feel like we prefer death to the prospect of continuing to live a life of torture.

Also the other weirder “I’m going to reincarnate as jesus” or “a comet will carry me to heaven” hallucinatory delusions I guess

Reply on news.ycombinator.com