I realize more all the time that I'm in the AI Skeptic camp. Every time I try to get an AI to do a simple task for me and I end up doing it myself I move further into that camp. I've got a whole list of failures in the accounts I or my employer pay for me. Confident python corrections that are definitely not correct, 14 attempts to have my publications on my CV reordered to meet the opposite requirements of my previous and current employer, and artwork that is hilariously awful. I fully expect everything "generated" from an AI to use more electricity than all of Panama to generate something that I will never be able to use (unless of course it blatantly stole it from some other source, in which case I can't use it anyway). I'm also convinced that several of my recent "peer" reviews were written by LLMs, but that's okay because they tend to be less critical of my work than my biological peers.
The goal is to do image recognition based AI on histological samples. Which would be the 100th time I've heard of someone trying to do this. There was a really cool company in Gaithersburg a while back that started up and shut down a while back, but they didn't have these LLM things, so maybe this is the real deal?
No comments:
Post a Comment