AI Reading for Thursday April 24
New LMarena is out with all the new models. o3 failed to dethrone Gemini 2.5 but came close. It's gotten confusing since there are so many different classes of models now. 4.1 has great JSON handling and simplifies agents pipelines.
OpenAI says it wants to make its upcoming open AI model best-in-class. Google has set a high bar with Gemma. - TechCrunch
OpenAI Forecasts Revenue Topping $125 Billion in 2029 - The Information
OpenAI gave ChatGPT Plus generous new usage limits. I want more deep research dammit. - TechRadar
VW will start testing robotaxis in US, aiming to deploy 'thousands' on Uber in 2026 - The Verge
Adobe and Figma integrate ChatGPT’s upgraded image generation model - The Verge
Adobe adds more image generators - The Verge
Also AI content protection watermarks - The Verge
Trump Signs Orders on AI Education, College Accreditation - Bloomberg
Startup Kollegio raises $2.8M for AI-powered college counseling service - SiliconANGLE
As ChatGPT scores B- in engineering, courses face shake-up - The Register
Google reveals sky-high Gemini usage numbers in antitrust case - Ars Technica
Google Blocked Motorola Use of Perplexity AI, Witness Says - Bloomberg
This text-to-voice model understands emotions like, "scream in terror" - Decrypt
The California bar exam was written with the help of AI. - Ars Technica
"Can you photoshop a FJÜNKSTÜHL into an Ikea ad? it's a barstool for depressive alcoholics with a built-in ashtray, a holder for a fifth of Jack, and a lap belt so you don't fall off"
Never trust a penguin in tap shoes with your lunch money. - Futurism
AQR Bets on Machine Learning as Cliff Asness Becomes AI Believer (gift link) - Bloomberg
Microsoft agents and the Frontier Firm - ZDNET
"I used the ‘cheat on everything’ AI tool and it didn’t help me cheat on anything" - The Verge
Improving Deep Learning With a Little Help From Physics - Quanta Magazine
Follow the latest AI headlines via SkynetAndChill.com on Bluesky