@arstechnica Do not base any decision you make on what an ai language model tells you. It is trained to replicate the style, tone, and other aesthetics of the text it is trained on. It only knows that these words often follow other words, not what those words mean or if what it is saying is true.
@arstechnica perhaps the public checking in with ChatGPT instead of a chat with a GP (ChatGP, eh? Eh!?), says something about the state of accessible healthcare.
@arstechnica I don't understand why it's so hard to get people to understand that LLMs are not actually AI and lack the most important part of that definition: an actual understanding. All they can do is put together denoising patterns. That can be incredibly useful in carefully controlled scenarios, but terrible for something like a general diagnostic...
Add comment