Zork

The eight large language models are able to put safeguards in by saying, okay, don't deny the Holocaust. But then if you want it to do that, we, you can get the note on my hand. It was easier than raw. As we've, as we've now seen, it's very easy to sort of bully an AI into having the opinion you want.

Play episode from 32:03

Transcript

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!

Get the app