
The 2022 AI Index
The Data Exchange with Ben Lorica
00:00
Is There a Way to Get Models to Not Generate Toxic Stuff?
Many actors, like open te ee ond others, have built sort of toxicity classifies into their systems so that they don't generate toxic stuff. This has created a counter reaction where some researchers are trying to train so called uncensored models as a response to this. Yes, we now have the same problems of hay pi ye es ot in be o.
Transcript
Play full episode