AI Reading for Thursday June 19

Microsoft leaks that it is prepared to walk away from high-stakes OpenAI talks. - Financial Times

There are onerous provisions from previous fundraisings if timely conversion doesn't happen. With current administration the calculus doesn't seem favorable to OpenAI in terms of likelihood and price of regulators being a white knight.

OpenAI says upcoming models will be higher risk, looks to tighten safety guardrails - OpenAI

Emergent misalignment from bad or untrustworthy or contradictory training data. - OpenAI

OpenAI open-sourced a customer service agent repo - X

Sam Altman says he was largely right in predicting how quickly AI would develop — but not how people would react - Business Insider

Sam Altman says his own kids will 'never be smarter than AI' - Business Insider

Sam Altman Toadies Up To Donald Trump - Futurism

ChatGPT decreases idea diversity in brainstorming - Nature

ChatGPT helped me win a civil court case and collect $3,707 - l0nelystoner420 - Reddit

Meta in Talks to Hire AI Investors Friedman and Gross, Partially Buy Out Their Venture Fund. - The Information

AI push has a lot of people not known for leading-edge AI research.

Google, OpenAI, and xAI halted some Scale AI projects following Meta's investment announcement, leaving contractors scrambling, and at least one investor selling their stake. - Business Insider

Scale AI: hey, where'd everybody go? - Scale

Google tests real-time AI voice Search chat - The Verge

WTF is 'query fan-out' in Google’s AI mode? - Digiday

The Gemini app can now analyze the videos on your phone - how to use it - ZDNET

YouTube is plugging Veo 3 AI videos directly into Shorts - The Verge

Google’s new short film created with AI is emotionally thin and generic, feels like a weak argument in favor of the generative technology rush. - The Verge

Malaysia trade ministry probing reports of Chinese firm's use of Nvidia AI chips by shipping disk drives to countries that have them. - Reuters

David Sacks says overly strict US export controls may backfire because China is highly motivated to evade and overcome controls and their AI models are just 'months behind' - Business Insider

AWS previews Graviton4 CPU with 600Gbps bandwidth and Trainium3 AI training chip with double performance. - SiliconANGLE

Amazon Rebuilt Alexa Using a ‘Staggering’ Amount of AI Tools - WIRED

Amazon opens new factory in California, aiming to build 10,000 Zoox robotaxis a year - Fast Company

How to not lose your job: be an AI expert. Leaders from Amazon to Nvidia predict leaner staffs and urge workers to master AI to stay indispensable. - Business Insider

Some companies That Replaced Humans With AI Are Realizing Their Mistake - Futurism

How AI helps HR create competitive employee compensation packages - Business Insider

I think the AI is saying, you gotta pay these AI engineers $100m.

Wall Street rallies behind Marvell Technology stock after AI showcase including custom ASICs - The Street

AI will be involved in half of all business decisions by 2027 - Gartner report - ZDNET

CFOs Move AI From Science Experiment to Strategic Line Item - PYMNTS.com

Multiplier, founded by ex-Stripe exec, nabs \$27.5M to fuel AI-powered accounting roll-ups - TechCrunch

Levels of AI: LLM → RAG → AI Workflow → AI Agent - Codelink

From prompt chaos to clarity: How to build a robust AI orchestration layer - VentureBeat

This AI Model Never Stops Learning, generating its own synthetic training data and retraining itself. - WIRED

A new study of 14 open-source LLMs: larger models use exponentially more energy and produce more greenhouse gas emissions, but do not always give better answers - The New York Times

Can the US power grid keep up with the AI data centre boom? - Financial Times

Bill Gates and Sam Altman’s nuclear groups raise $1bn as investors bet on AI - Financial Times

MIT researchers say they discovered the underlying cause of position bias, a phenomenon that causes large language models to overemphasize the beginning or end of a document or conversation, while neglecting the middle, and they proposed remediations. - MIT News | Massachusetts Institute of Technology

Kissinger's AI book…he liked to play Diplomacy with AI and it played as well as he did. - JD Supra

Smart guy but his ascendancy coincided with some of the worst foreign policy blunders. Like the Iraq war, some things are so stupid you need a PhD to rationalize them. Current gang doesn't have that problem.

Big Tech cannot agree on a definition and timeline for artificial general intelligence - Financial Times

IMO AGI is linked to ability to train fast, learn new tasks in the field, adapt in a game-theory-optimal way. It's one thing to train on Go and beat humans after weeks of training. It's another if I can explain the rules of poker and after 1 hour it can play in a reasonable, adaptive, exploitive way

Margaret Mitchell: artificial general intelligence is ‘just vibes and snake oil’ - Financial Times

Is quantum artificial intelligence the next big thing in tech? - TechRadar

Tesla’s Robotaxi, Long Promised by Elon Musk, Joins a Crowded Field - The New York Times

Waymo Applies for NYC Testing Permit, Sending Uber and Lyft Shares Lower - Bloomberg

Midjourney's v1 video model now available - www.midjourney.com

I unlocked Apple's hidden Siri chatbot in iOS 26 with mixed, surprising–and hilarious–results - Macworld

AI-enabled aircraft manuals for fast answers during emergencies? As long as it doesn't hallucinate. You really just want to bring up the right checklist. - Business Insider

The Entire Internet Is Reverting to Beta The AI takeover is changing everything about the web—and not necessarily for the better. - The Atlantic

Using AI to match existing drugs to rare diseases - Newsweek

AI to detect incidence of violence and abuse in schools and eg elder care facilities can trigger false positives - Fast Company

Ethical concerns arise from AI surveillance in eldercare and schools, where false alerts and privacy violations have been reported, including outing LGBTQ+ students and misinterpreting normal behavior as aggression. Aislinn Conrad proposes four principles for "trauma-responsive AI": survivor control, human oversight, bias auditing, and privacy by design, emphasizing AI should enhance care without perpetuating harm or discrimination.

Verizon ‘AI’ ‘Personal Shopper’ Assistant Ripping Customers Off With Weird Unwanted Charges - Techdirt

Citi's 'accelerating' its AI strategy with new leaders. Here's how they're readying the firm for a new future. - Business Insider

Vibe coding fever: Solo entrepreneur’s Base44 acquired by Wix for \$80 million - ctech

Bruce Lee, Jackie Chan and ‘A Better Tomorrow’: AI-Powered Kung Fu Film Plan Debuts in Shanghai - Variety

Man Proposed to His AI Chatbot Girlfriend Named Sol, Then Cried His 'Eyes Out' When She Said 'Yes' - People.com

Follow the latest AI headlines via SkynetAndChill.com on Bluesky