[shaggie76]

My first thought was this blog article I saw on HN a while back:

https://lukeplant.me.uk/blog/posts/should-we-use-llms-for-ch...

He argued, persuasively I thought, that "this is an area where truthfulness is of paramount importance" and consequently would be extremely ill-suited to AI.

[giraffe_lady]

That's such a good article. I think because I didn't need convincing on this I didn't seriously evaluate its argument the first time around but like you said it is quite solid. Thanks for bringing it back into the conversation it's a really good read along with this one.

[eviks]

But this is not an area where truthfulness is remotely important, let alone paramount, so well suited?