So, I guess they run out of data to train on ...

I wonder on how much they can rely on the data and what kind of "knowledge" they can extract. I never give feedback and most time (let's say 5 out of 6) the result cc produce it simply wrong. How can they know the result is valuable or not?

▲

debesyla 4 days ago | parent | next [-]

Maybe they can use same method as Google does - if user clicked a link and didn't try to search again, it can assume the link had intended result.

So your silence can be used as a warmish signal that you were satisfied. (...or not. Depends on your usage fingerprint.)

	▲	rurp 4 days ago \| parent [-]
		I expect that's a very weak signal. When I ask a question and get a completely wrong answer from Claude I usually drop the chat and look elsewhere.

▲

jlarocco 4 days ago | parent | prev [-]

How can they know anything they train on is valuable?

At the end of the day it doesn't matter. You got the wrong answer and didn't complain, so why would they care?

	▲	vb-8448 4 days ago \| parent [-]
		In general, I think any human generated content in pre-2022 is valuable because someone did some kind of validation (think about stack overflow answer with user confirming that a specific answer fixed their problem). If they start to feed the next model with LLM generated crap, the overall performance will drop and instead of getting a useful answer 1 of 5 it will be 1 of 10(?) and probably a lot of us will cancel the subscription ... so in the end I think it matters.