SERI 2022: Timelines for Transformative AI and Language Model Alignment | Ajeya Cotra
EA Talks
00:00
Is the Terminator Really Real?
Experts and leaders in ti companies not engaging the same sort of problem as you are. Even language predictors might well become internally consequentialist, he says. And a pursuing those golls really hard might lead to a sort of power seeking r or a i tape over scenario a. He is more worried about language models that have been further fine tuned with reinforcement learning than language models that are pure predictions.
Transcript
Play full episode
