Not everyone who uses ChatGPT gets advice on harming or killing themselves, but some people sure do. And some act on it.
Just as we don't necessarily want to eliminate a technology because of a small percentage of bad cases, we shouldn't push a technology just because of a small percentage of good anecdotes.
This genie isn't going back in the bottle but I sure hope it ends up more useful than harmful. Which now that I write it down is kind of the motto of modern LLM companies.
Just as we don't necessarily want to eliminate a technology because of a small percentage of bad cases, we shouldn't push a technology just because of a small percentage of good anecdotes.
This genie isn't going back in the bottle but I sure hope it ends up more useful than harmful. Which now that I write it down is kind of the motto of modern LLM companies.