
#319 - The Digital Multiverse
Making Sense with Sam Harris - Subscriber Content
00:00
The Importance of Not Making Errors
In the presence of a chatbot that is simply not making errors, how are you going to detect its failure of reasoning? Yeah. In terms of content generation, you're okay. It's when you start asking it to explain and elaborate and construct a coherent worldview narrative... That's when you get into trouble. And I can't think of any case in which I haven't been able to trip up an AI by asking it to elaborate repeatedly. So AIs can crush chess and go because you've defined the landscape very clearly. They could probably be pretty good customer service agents. But again, these are things in which already customer service agents are reading off of scripts.
Transcript
Play full episode