AI Reading for Monday August 11

GPT-5's coding skills questioned - ZDNET

Sam Altman says increasing reasoning limits for Plus users, maybe giving them a couple of pro queries per month, other stuff - X

Altman: Gen Z is the luckiest generation due to AI. Call no man lucky until he is dead - Fortune

GPT-5 rolls out multiple personalities as research preview - Tom's Guide

GPT-5 hallucinates less but it is less creative and sycophantic, you might have to prompt it differently, it's just very different and a lot of people don't like it. - Reddit

Gary Marcus says GPT-5 release was underwhelming. Also there's a lot of hype. Also current algos don't generalize out of sample, and scaling doesn't get you to AGI. In other words, all the obvious stuff. - Gary Marcus Substack

I used ChatGPT as my career coach for the week. It gave me some good advice, but couldn't handle my meltdown. - Business Insider

Sam Altman says he's 'uneasy' about people trusting their biggest life decisions to ChatGPT - Business Insider

Berger's three laws of chatbotics:

1. The bot should never express emotions.

2. The bot should never praise the user.

3. The bot should never say it understands the user's mental state.

There's no use case for these things, and they're obviously problematic

Google fixes 'Marvin the Paranoid Android' mode - The Register

Is the A.I. Boom Turning Into an A.I. Bubble? - The New Yorker

Census counts 498 AI unicorns - CNBC

A 23-year-old former OpenAI researcher who wrote a viral manifesto quickly amassed more than $1.5 billion for a firm he called a “brain trust on AI.” - The Wall Street Journal

The AI ‘Big Short?’ Investors are starting to flee stocks that are seen to be vulnerable to AI disruption - Bloomberg

AI is already powerful enough to replace millions of jobs. The only thing stopping the mass layoffs isn't technology. It's that no CEO wants to be the first one to face the political firestorm. - Gizmodo

Recessions are when everyone cuts, if you are AI exposed, next one might be a doozy

Goodbye, $165,000 Tech Jobs. Coders Seek Work at Chipotle. - The New York Times

AI decimates entry-level dev jobs, boot camps, lengthens junior job searches - South China Morning Post

Workers flock to AI education programs - The Washington Post

AI is coming for (some) finance jobs - Financial Times

What the man on the street is doing with AI - The New York Times

What people in a restaurant do with AI - Gizmodo

Figma’s CEO said AI enables generalists, blurs roles across product and engineering. - Business Insider

Greg Brockman says engineers need technical humility, ability to listen deeply and collaborate - Business Insider

Also, engineers who can innovate from first principles on things no one did before, build fast from the ground up - Business Insider

9 AI business models - PwC

Researchers report that masked diffusion language models (DLMs) "are super data learners" - Notion

Sebastian Raschka: From GPT-2 to gpt-oss: Analyzing the Architectural Advances And How They Stack Up Against Qwen3 - Sebastian Raschka's Magazine

Are nukes the only thing that can provide all the power the AI revolution needs? Big Techs are buying. - GIS Reports

Nvidia and AMD to pay 15% of China chip sale revenues to US government in exchange for export license, under no apparent legal authority - Financial Times

Manus and Benchmark: AI deal upsets both China and the US. Agent start-up came under fire from Beijing for moving to Singapore, as Washington scrutinized its Silicon Valley backer - Financial Times

Enterprises deploy AI agents for cyber defense, automating security workflows while facing evolving threats. - CNBC

Can AI 'Defeat' Authentication? Depends on Who You Ask - The New Stack

Boston Public Library partners with Harvard and OpenAI to use AI to digitize and increase access to a vast historic archive - NPR

Apple’s Upcoming AI Voice Control Will Change How People Use iPhones (gift link) - Bloomberg

Amazon’s Alexa+ troubles justify Apple’s slow pace with next-gen Siri - 9to5Mac

"In addition to A.I. deepfakes, there are more risks to photo posting — like potentially exposing young people’s sensitive personal information to bad actors — that may discourage parents." - The New York Times

This by @zey.bsky.social ends with a nod to a real risk- synthetic media is likely to give authoritarians even more of an excuse to step in to surveil and enforce 'reality' in various ways. - The New York Times

Sex is getting scrubbed from the internet, but a billionaire can sell you deepfake AI nudes of Taylor Swift - The Verge

Here’s how to spot AI writing, according to Wikipedia editors - THE DECODER

AI oral cancer screenings with 95.9% AUC which is pretty pretty good - Nature

measures overall quality of a classification model which you can then tune for better selectivity or sensitivity depending on cost of false positives v. false negatives.

Coral Protocol allows many small agents to cooperate using blockchain (hype singularity warning) - Hacker Noon

Could a super-mixture-of-agents combining the top open-source models beat the best closed-source models? - Hacker Noon

As AI becomes more pervasive, we’re becoming more forgiving when algorithms mess up - The Conversation

Just say no to digital resurrection - the Guardian

In New Dating App, You Talk to an AI Matchmaker Who Sets You Up on Dates With Strangers - Futurism

"The road to hell is really fun" say porn addicts - WIRED

Philadelphia SEPTA deploys AI cameras to curb bus-lane violations - Metro Philadelphia

Follow the latest AI headlines via SkynetAndChill.com on Bluesky