> I see remarks like this a lot, and I don't know what to do with them.
There's no "reasoning loop" built into LLMs yet. Keyword, yet. For now we're left with single-shot answers from "memory" rather than a reasoning loop akin to what a human would do, which is read the docs, try some stuff out, discover that what you're asking for is impossible, and then telling you that it isn't possible.
There's no "reasoning loop" built into LLMs yet. Keyword, yet. For now we're left with single-shot answers from "memory" rather than a reasoning loop akin to what a human would do, which is read the docs, try some stuff out, discover that what you're asking for is impossible, and then telling you that it isn't possible.