Hacker Read top | best | new | newcomments | leaders | about | bookmarklet login

Well, we use humans to learn from mistakes in both cases and improve the technology.

The big difference is a mistake in a car is more likely to cause injury than a misdelivered letter.



sort by: page size:

Humans make mistakes too, but the kinds of mistakes tend to be different.

Human mistakes also happen. How frequently do humans mess up compared chatGPT becomes an interesting question.

Devil's Advocate: What makes this any different than human error?

Humans occasionally make mistakes, including expert ones.

Not by accident, by carelessness. There's a difference.

Humans also make mistakes, all the time.

Sometimes a mistake is a decision under the assumption that the people intended to use this are smarter / more careful than they are.

Humans aren't perfectly informed and sometimes they make mistakes.

Of course. My point is that people find ways to screw up, regardless of the underlying technology.

Yes, humans make mistakes, and we improve by creating strong disincentives for mistakes.

Human error: somebody sometimes did something they were not supposed to do. But they didn't know so they've done it anyways.

Hey, well said.

Fuck-ups happen. The biggest difference is how you deal with it.


But how is it deferent from humans? They make mistakes too! All that matters is rate of mistakes.

It's also way easier, and much less responsibility, to miss a mistake than to make one.

"Oh, yeah, that was a fuckup but the AI did it, not me! What, I was supposed to catch it? Well, I blame the AI!"


There is a moral difference between accidental mistake and deliberate action.

I think there is some nuance here; not all screw ups are equal, nor are all ways of correcting mistakes equal.

Are 2 errors better than 35,000 human errors?

One thing is making a certain mistake because a choice based on erroneous or insufficient data has gone wrong (No matter how great the AI, this will always be an issue), and a very different one is making a mistake even with complete and correct data, something that humans are prone to.

Yes. “People make mistakes too” isn’t a very useful idea because the failure modes of people and language models are very different.
next

Legal | privacy