Nimitz14

Great question. I think these people are good at spinning narratives _and_ the people above are not doing the work they should be doing to check whether the narrative checks out: I think good leaders (and every manager below) should be doing inverse skip and skip-skip meetings often enough to have a good sense of whether what their reports are saying is accurate. But most dont

You have some fundamental misunderstandings on how LLMs work.

You’re not getting it. The probabilities do not change at all. The only change is given some probabilities there is a deterministic method for determining which symbol was sampled from that distribution. The distribution or sampling process itself is not modified.

Uh isn’t google known to have the best tooling in the world

Ai slop

He said it nudges it to be more deterministic. Your comment is not correct.

This is mostly slop. But you may be directionally correct

Really really cool. If they can serve this efficiently it would disrupt a lot of things.

I think your own math leads to the conclusion the public apis are not serving models of that size. They couldn’t afford to

3/4 are actually negative value for most customers

Your answer presumes we cannot hold people accountable. I think that is incorrect.

Most of these errors will not be meaningful. Real speech is full of ambiguities. 3% is low

I think you're bang on the money fwiw. But also worth mentioning that it's OK to ask rather than trying to predict and feeling that having to ask means failure