Axios used AI to fake an opinion poll

https://sh.itjust.works/post/58082125

Holy fuck. It can simulate large samplings or it can just hallucinate some nonsensical BS that completely misinterprets the data it gathers in order to agree with the phrasing of the person who created the prompt.

Do the majority of people trust their doctors and nurses? Maybe. Or, maybe it depends on the context of the question.

Do I trust my doctors and nurses are a better source of information than random internet advice and AI generated slop? I would hope so.

Do I trust that the American healthcare system is set up to prioritize the health and well-being of the patient over maximizing profits and forcing healthcare workers to adhere to standardized time allotments of 10 to 15 minutes for every patient interaction regardless of the individual case? Absolutely not.

Yep. AI, such as it is, can be useful for some things. I had had a song kinda stuck in my head for years, couldn’t remember the lyrics, artist, or song name - only that it was in french and done stuff from the video. ChatGPT worked it out quickly. On the other hand, I was using it to look up definitions for a legal case I’m kinda involved in that hasn’t really happened before - I know I shouldn’t, but I’m not a lawyer! - and I found it was using my own case which has not been decided as evidence.

So yeah, they can be impressive but you can’t trust these programmes with anything important.