←back to thread

323 points steerlabs | 1 comments | | HN request time: 0s | source
Show context
jqpabc123 ◴[] No.46153440[source]
We are trying to fix probability with more probability. That is a losing game.

Thanks for pointing out the elephant in the room with LLMs.

The basic design is non-deterministic. Trying to extract "facts" or "truth" or "accuracy" is an exercise in futility.

replies(17): >>46155764 #>>46191721 #>>46191867 #>>46191871 #>>46191893 #>>46191910 #>>46191973 #>>46191987 #>>46192152 #>>46192471 #>>46192526 #>>46192557 #>>46192939 #>>46193456 #>>46194206 #>>46194503 #>>46194518 #
HarHarVeryFunny ◴[] No.46191893[source]
The factuality problem with LLMs isn't because they are non-deterministic or statistically based, but simply because they operate at the level of words, not facts. They are language models.

You can't blame an LLM for getting the facts wrong, or hallucinating, when by design they don't even attempt to store facts in the first place. All they store are language statistics, boiling down to "with preceding context X, most statistically likely next words are A, B or C". The LLM wasn't designed to know or care that outputting "B" would represent a lie or hallucination, just that it's a statistically plausible potential next word.

replies(7): >>46192027 #>>46192141 #>>46192198 #>>46192246 #>>46193031 #>>46193526 #>>46194287 #
toddmorey ◴[] No.46192027[source]
Yeah, that’s very well put. They don’t store black-and-white they store billions of grays. This is why tool use for research and grounding has been so transformative.
replies(1): >>46192514 #
1. therealpygon ◴[] No.46192514[source]
Definitely, and hence the reason that structuring requests/responses and providing examples for smaller atomic units of work seem to have quite a significant effect on the accuracy of the output (not factuality, but more accurate to the patterns that were emphasized in the preceding prompt).

I just wish we could more efficiently ”prime” a pre-defined latent context window instead of hoping for cache hits.