AI Reading for Tuesday March 11
Had to skip Monday but we’re back baby!
The Quest for A.I. ‘Scientific Superintelligence’ and a century of scientific progress in a decade - NY Times
AGI is now a dinner table topic. - MIT Technology Review
The Einstein AI model - Thomas Wolf
I don't know what Einstein AI looks like but this ain't it. Need better representation and ability to learn on the fly. Then you have AI that can adapt in game-theory-optimal ways, and that’s what I would consider AGI.
Can AI stir-fry? - The New Yorker
There's crazy hype and people using AI for things it's no good at, or just silly. But I'm more in the camp of, we are screwed because AI is pretty good and getting better fast, and humans won't be able to adapt quickly enough to avoid disruption, social strife, and war.
What the Dot-Com Bust Can Tell Us About Today’s AI Boom - WSJ
Manus mania - The Register
Google Removes 'Diversity' and 'Equity' From Responsible AI Page - MEDIANAMA
Stanford AI benchmarks measure bias, ability to understand context and differences between people. - MIT Technology Review
(can't have that)
OpenAI in huge CoreWeave deal giving them equity ahead of IPO - TechCrunch
Can AI help executives make better decisions faster, with better data from more varied sources? - TechRadar
Apple's friendly home hub postponed due to Siri delays - TechRadar
Rumored to make expressive gestures like the Pixar lamp? - YouTube
CJR finds that most of the time AI doesn't cite news stories correctly and link back - CJR
AI skills command large wage premiums - CIO Dive
Silicon photonics chip company calls itself Celestial AI. Is this really an AI chip company though? - Bloomberg
AI in vineyards. I dunno, if a tractor operator is overseeing a fleet of self-driving tractors, someone is losing their job. - Financial Post
With reasoning models and increased adoption, inference compute demand is taking off. If Nvidia's chips have a larger advantage in training, it would be to their advantage to segment the market/price-discriminate, and release a cheaper inference chip. But that's not the way they are going.
Meta begins testing its first in-house AI training chip - Reuters
Gemini Extensions Are Now ‘Apps’ and Powered By Gemini 2.0 Flash Thinking Experimental Model - Gadgets 360
Let's ask it to make a list of songs about coffee and espresso - Google Docs
meh
Amazon wants launch drone trials in the UK - FT
FT looks at Amazon's robots - FT
Bluesky is working on a consent framework to let people decide over how their content is used to train AI - TechCrunch
'Ban AI for political use,' says co-creator of Trump Gaza AI video - The Globe and Mail
AWS AI team including Bedrock is in major reorg mode - Business Insider
The confusing story of GSAI - Inc
Follow the latest AI headlines via SkynetAndChill.com on Bluesky