AI Alignment

The difficult thing is right like you train the model right and nowadays most of the models are trained on or it's just what's on the internet. But increasingly it will be enterprises, you know specific data tkks, you know order history For all their users right and how they speak to the staff, maybe that's how they train them. It's kind of like alignment is basically giving a moral compass or trying to give a moral compass to an AI because if you train it on the internet, you know significant portion of the data will be pro x or y or z Which we as society find? Reprehensible, but there are people on the internet who post about it.

Transcript

Play full episode

Transcript

Episode notes

The AI-powered Podcast Player

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

Get the app