AI Stuff I Saw Today, Friday 2024-03-08
Juicy details on Mira Murati's role in Altman saga. It can be a fine line between strong company culture and a cult, or between strong leadership and cult of personality or toxicity. - New York Times
President Biden Calls for Ban on AI Voice Impersonations During State of the Union - Variety
America runs out of power capacity - Washington Post
Claude-3 Opus approaches latest GPT-4 on Chatbot Arena, passes Google Gemini Pro but not Ultra / Advanced which is still unrated. - HuggingFace
Inflection launches new model, said to nearly match GPT-4. - VentureBeat
Oh hey there's a Chatbot Arena paper now too. - HuggingFace
Jailbreaking AI with ASCII art - Tom's Hardware
Human ability to detect AI-generated pictures only slightly better than chance. - TechXplore
Reddit debates images above. - BGR
Reddit rolls out AI bouncers, after human moderators revolted, and ahead of expected IPO - The Register
8GB Gemini Nano will run on phones in theory, but not currently in practice. - BGR
AI Incident Database launched to help track AI risks - The Register
NIST, a leading Federal agency for AI policy has staff revolting over an 'effective altruist' appointee - VentureBeat
NIST also has a leaky roof. - Washington Post
Using AI for resume screening might not work very well, risks biases, might fall on the wrong side of some hiring laws - Bloomberg
Some more dataviz - Bloomberg
Layoffs surged in February to highest level since 2009 — and AI is a big reason - New York Post
Better AI might make prompt engineers irrelevant - Business Insider
Train/fine-tune a 70b parameter model on your desktop with 2 4090s - Answer.AI
Robots gonna be outrunning you and doing that viral Rasputin dance any day now - Popular Science
One Month With Apple Vision Pro: In the Air, on a Train and…in a Drawer - WSJ