Sandhini Agarwal: We did find that it produced sure undesired outputs, but they had been all things that GPT-3.5 also generates. So concerning danger, as a research preview—for the reason that that’s what it had https://youtu.be/aGubCnA5Qyk