News for hackers. Stuff that matters.

A comment on GPT-4. Parent comment

Re:GPT-4 by acc_297 on Tuesday March 14, 2023 @01:52PM (#35155537)
Yeah looking at the responses they include without using a safety layer it’s pretty clear that the underlying unfiltered model assigns quite a bit of truth to 4chan-esque ideals and values

It’s an open question how much of this makes it through the safety layer like if asked to interview job candidates would these undesired biases make it through or are they caught along the way

[ Parent | on HN ]
Re:GPT-4 by AndrewKemendo on Tuesday March 14, 2023 @03:00PM (#35156768)
It means growth is bottlenecked by the terrible data

So the linearly growing safeguards will either stifle the growth of the underlying models

or, more likely

After a certain point people throw their hands up about the guard rails because integrations have obviated people who understand the system and they have no idea how to unwind it

[ Parent | on HN ]
Re:GPT-4 by subsistence234 on Tuesday March 14, 2023 @06:20PM (#35159872)
we need to remove empirical data and stats from the training data, to prevent the AI from noticing the wrong things.
[ Parent | on HN ]
Re:GPT-4 by thomastjeffery on Tuesday March 14, 2023 @09:14PM (#35162329)
But what can go in their place?
[ Parent | on HN ]