4 ms·
Is there a risk of it hallucinating what new features could do instead of basing information on having seen them before?
by grenoire 3y ago
Is there a risk of it hallucinating what new features could do instead of basing information on having seen them before?
- RadiozRadioz 3y agoYes, good point, I suppose we would have to weigh the probability of that against it fudging a number. I would assume inventing a new feature is harder, but who's to say; all this LLM stuff comes down to probability.