Oh course Tucker makes some common mistakes about AI. Like saying its doing things autonomously and lying. Every story I've seen about things that make these claims are false or misleading.

By doing thing autonomously I believe he's referring to the "testing" being done by some of these AI companies that say the AI is doing things that make no sense. I'm skeptical about these stories as they sound like self serving hype to get investor's attention.

From what I've learned about LLMs what is going on is the chatbot is creating responses the algorithm calculates would be an expected response. There is no will or intent. This has yet to be shown.

And LLMs know nothing. They are prediction machines that generate text that statistically would be expected based on their training. It might seem like a subtle difference but I respected the Bishop's humility on the subject. I wish Tucker would show more restraint when he talks about technology he doesn't understand. His past statements on Signal for example. I don't trust Tucker.