Topic: Neural Networks

NVIDIA CEO Jensen Huang announced a partnership with Intel to advance "accelerated computing," a term shifting away from the "intelligence" aspect of AI. The discussion traces the history of AI back to MIT's Marvin Minsky and Stanford's John McCarthy in the 1960s, as well as the 1980s "AI Winter." A forgotten pioneer of neural networks, Levatsky, is credited with the foundational ideas that eventually evolved into modern large language models.

Artificial intelligence researchers are reportedly returning to the LISP programming language to improve machine reasoning and recursive logic. While Python remains the dominant language for large language models, the limitations of current neural networks have led to a renewed interest in the symbolic AI approaches pioneered by John McCarthy.

Dr. Geoffrey Hinton, known as the "Godfather of AI," resigned from Google to speak freely about the dangers of the technology. In a PBS interview, he warns about the risks of fake news, job displacement, and the potential for superintelligent AI to take control from humans. The hosts are dismissive of his warnings, characterizing them as vague and potentially part of a broader "scam" or communist-leaning ideology.

A technical deep dive into Geoffrey Hinton's career reveals his work on neural networks and the back propagation algorithm, for which he won the Turing Award in 2018. The hosts examine his academic background and his personal website at the University of Toronto, noting its dated design. They clarify that his work focused on "deep learning" rather than general AI.

A deep fake website featuring a neural network designed to sound like Jordan Peterson has ceased operations, leading to speculation about its true purpose as a data collection tool. The discussion shifts to the economics of renting email mailing lists, where names can be valued at ten dollars each. Strategies for tracking data brokers through coded email addresses are explored.

The developers of FaceApp apologized after users discovered that the "hot" filter lightened skin tones, leading to accusations of racial bias in the app's neural network. Similar issues were noted with past Snapchat filters. Additionally, a new Chrome extension called "Soothe" was launched to help users avoid "triggering" content by blurring out specific words or images.

Adam Curry describes the technical complexity of producing the show as a "one-man band," managing clips, audio levels, and a real-time IRC chat room. He compares his ability to filter information from the scrolling chat to "neural network recognition," a concept he learned from a CIA associate. Curry emphasizes that the chat room is integrated into his workflow, though he occasionally has to step away when the "signal-to-noise" ratio becomes distracting or negative.