Daily AI Reading for Wednesday April 3

A thinking cap from UT-Austin researchers lets users control a computer and play Mario Kart with their brain, after some automated calibration. - TechXplore

---

Google hired OpenAI's former head of developer relations, a notable win in the war for talent. - Business Insider

---

Stability AI ran out of money, was on track to spend $99M annually on $11M in revenues.

---

Dall-E now lets you select an area of a created image and ask it for changes. - OpenaI

---

Using voice fingerprinting to identify customers is a terribly unsafe idea, but some big financial institutions are doing it. - WSJ

---

AI is better at persuading people in online debates than humans. - The Reg

---

Can an AI-first IDE provide better integration, make devs more productive, make them cyborgs instead of centaurs? (IYKYK) - Twitter

---

Spatial personas let you use your Apple Vision Pro to sit around a room with 4 other remote people and interact with them. Remember, it’s not virtual reality, it’s spatial computing! And they are not avatars, they are spatial personas! This isn’t Second Life, people. - Inverse

---

What qualifies as AI? Chatbots? translation? directions? Users and experts share views. - Washington Post

---

How AI depicts queer people. - WIRED

---

A king of NYC commercial real estate is moving into AI data-driven real estate sales. - Bisnow

---

Artifact AI-powered news startup shuttered, tech sold to Yahoo! News. AI news is a fool's game for sure. - ZDNET

---

Cloud providers compete to give startups AI and cloud service credits, AWS gives YC startups $500K.

---

Amazon moves to mobile scanners, ends Just Walk Out in its stores, which turns out to have been powered not by AI but by 1000 offshore workers in India. - Futurism

---

Jailbreaking LLMs by acting innocent, winning their confidence, and then, blam! - TechCrunch

---

AI gun detectors have been tried and found ineffective. Seems like a gimmick and a poteintial invitation to arbitrary cop searches and a slippery slope. But, as devil's advocate, game theory implies an optimal level of random inspections and any real improvement over the base rate is a win. - The Verge

---

Judge bans AI-enhanced-cough-manufactured evidence, but how exactly do you prove evidence was not AI-manufactured? - Gizmodo

---

Hollywood mobilizes to combat deepfakes. May only our legally sanctioned world of make believe survive! - LA Times

---

Oren Etzioni of Allen Institute releases deepfake detection tools. - NYT

Once it goes viral, if it's juicy enough and truthy enough nobody will care if some egghead debunked it. - The Mercury News

---

AI-generated influencer to host 'Survivor' segment in Spain - New York Post

---

AI maven Ethan Mollick, author of 'Cointelligence', on Ezra Klein podcast - NYT

---

Experiment with your own $250 robot arm. - GitHub

---

Follow the latest AI headlines via SkynetAndChill.com on Bluesky