• Aceticon@lemmy.dbzer0.com
    link
    fedilink
    English
    arrow-up
    3
    ·
    1 day ago

    The AI doesn’t “deem” anything.

    The details being left out is just the product of, statistically, details being a lot rarer in the text used to train the AI than broad strokes overviews or even just they not always being in a consistent place in different texts and/or kept separate from overview text so the LLM, following the highest probability line along the N-dimensional token-relationship space encoded in its trained model whilst it generates the “explain quantum physics” text, never or almost never goes into the point clouds were the details are.