AI Reading for Wednesday March 5

Judge denies Musk's request for injunction against OpenAI's for-profit conversion, says other claims can proceed. - CNBC

---

DOGE staffer's GitHub (now private) had a project that looks suspiciously like an AI agent to evaluate Federal employees for possible firing using their work history, public/social profiles etc. - Fast Company

---

DOGE AI and 'break first, fix later' approach worries some observers - CNN

Just learn to code, all you loser journos!

---

Sesame startles with human-sounding conversations - Ars Technica

---

McDonalds brings more sensors and AI to stores for real-time monitoring of operations. - WSJ

---

Turing award goes to Sutton and Barto who are the godfathers of reinforcement learning. - ACM

---

AI movie studio wants to create 30 $500k movies in 4 years. - TheWrap

---

Moving towards the Star Trek computer - Hacker Noon

---

AI isn't slowing down Google search yet, but maybe we overestimate the effect in the short run and underestimate the effect in the long run. - Business Insider

---

Surprise, AI counterpoints to LA Times opinions cause problems - Nieman Lab

Cool, op-ed opinions can now be self-rebutting using AI.

---

Should AI sidestep complex ethical problems, or let people have discussions about them, and even surface takes that some might regard as terrible? - The Verge

---

Thousands of pedophiles are using jail-broken AI character chatbots to roleplay sexually assaulting minors - Fast Company

---

An interview with MIT’s David Autor - Sloan Review

"We can have a Taylor Swift of popular entertainment who can entertain 2 billion people, but we can’t have a Taylor Swift of oncology or of medicine in general who can take care of 2 billion people."

There's stuff the doctor does that cannot go into an AI but 50% of what they spend time on maybe could go into an AI that serves 2 billion people?

---

Scientific American has a special edition on AI - Scientific American

---

Will a future superintelligent AI punish people who fought against bringing it into existence? If everyone had to decide a priori in some weird Rawlsian thought experiment that defeated causality, maybe? does that show some limits to Rawlsian thought experiments? - LessWrong

---

CoreWeave in Talks to Buy AI Startup Weights & Biases (leading platform for AI model evals) For Around $1.7 Billion - The Information

---

Krane's AGIX fund has private Anthropic stake - Yahoo

---

Anthropic's product chief said DeepSeek had 'almost no impact' on the $61.5 billion company's strategy - Business Insider

definitely impacts the competitive landscape if there is an open-source lab releasing stuff that is close to or matching the leader.

---

Google rolling out Gemini features similar to OpenAI's Advanced Voice Mode with Video to users later this month - Tom's Guide

---

AI for soccer scouting - SiliconANGLE

---

Software companies going all-in on agents - Business Insider

---

Columbia sophomore makes AI tool to assist technical coding interviews, gets offers from 3 big tech employers, posts video on YouTube. Amazon sees video, complains to career services office. Student presumably about to raise a seed round after dropping out. Go Lions! - Gizmodo

Our boy is spicy. - Xitter

For real tho, using the process and YouTube to go viral is a bit scrappy/shameless

---

YouTube warns creators an AI-generated deepfake video of its CEO is being used for phishing scams - The Verge

---

Salesforce launches agent marketplace - VentureBeat

---

Amazon's AWS forms new group focused on agentic AI - Reuters

---

Asking ChatGPT about trauma increases its anxiety, mindfulness exercises decrease it. - The Register

---

So it begins … they are testing our defenses - Businesskorea

Praise Kier, Deep Research, what is the deal with Macrodata Refinement? - ChatGPT

---

Follow the latest AI headlines via SkynetAndChill.com on Bluesky