Any thoughts on carefully micromanaged AI, in the hands of experienced researchers with in-depth knowledge of their respective fields?

February 18, 2026 - 19:31 UTC

1
0
0
Experienced researchers don't need a machine to output answer-shaped objects. LLMs can never do anything but string words together probabilistically. They don't synthesize new knowledge, and can't be stopped from making up nonsense no matter how well curated. They're not even a good search engine.
1
0
8
Not for writing papers, just as research tools, but if they can't be good search engines with proper controls in place, that throws that idea out of the window.
1
0
0