minus-squareMereTit@lemmy.worldtoSingularity | Artificial Intelligence (ai), Technology & Futurology@lemmy.fmhy.ml•OpenAI: Introducing SuperAlignment (blogpost from 5.07.2023)linkfedilinkEnglisharrow-up2arrow-down1·1 year agoThey say that they are deliberately training misaligned models to test on… what does that mean for safety? linkfedilink
They say that they are deliberately training misaligned models to test on… what does that mean for safety?