Aligning Super Intelligence & The Open Web's Uncertain Future + GPT-4 General Availability | EP22
In Episode 22 we cover GPT-4 general availability and the implications to millions of developers. We discuss George Hotz revealing how GPT-4 works & Alignment of Super Intelligence. Will OpenAI's Super Intelligence succeed? We also cover the implications of AI agents on the open web, OpenAI disabling web browsing in ChatGPT temporarily to protect content creators and discover Jolly Roger's AI time wasting for service to trick scammers.
If you like this podcast please consider subscribing, liking and leaving us a comment. We really appreciate your support of the show.
00:00 - Cold open
00:27 - GPT-4 API General Availability for Developers
06:40 - ChatGPT Code Interrupter
15:30 - ChatGPT as Work Assistant: Code Interrupter + Vision
16:42 - ChatGPT web traffic down by 10% & thoughts on single agent Vs AI everywhere
21:03 - George Hotz leaking GPT-4 is a 16 way mixture model
26:50 - Is AI alignment just a human alignment problem?
27:57 - Is alignment making GPT-4 and ChatGPT worse? Discuss on AI alignment
38:18 - Is the best defense against super intelligent AI giving everyone super intelligent AI?
40:55 - Is the Open Web Doomed? AI's impact on the Open Web
54:42 - OpenAI's 20% compute power to "Superalignment"
1:00:29 - OpenAI now sends email threats
1:03:12 - Waste Scammers Time with This AI Voice Tool