
The Potentially Dangerous Future of AI
WSJ Opinion: Free Expression
00:00
The Progress of AI
The way these systems work is that they learn from data, they learn from examples. They discover their own knowledge in a way that's not in a form that humans can easily kind of analyze. Instead, they're more like their brains. And so these AI systems in particular, like chat GPT are trained on a large fraction of everything that can be available on the internet in terms of texts. That's many orders of magnitude bigger than any human could potentially read in their lifetime. But again, if you were just in pretty simple layman's terms, explain how it works.
Play episode from 05:42
Transcript


