AI Reading for Monday August 4

Inside OpenAI’s quest to make AI do anything for you - TechCrunch

Sam Altman says 'bear with us' through capacity constraints, suggesting GPT-5 is imminent - X

Sam Altman teases GPT-5, shares screenshot of it recommending Pantheon as most thought-provoking TV show about AI - Business Insider

Reviewing what people speculate is OpenAI's open source model, which is available for testing under a pseudonym - X

Will data centers crash the economy? - Noahpinion Blog

In Trump’s Washington, Palantir wins big with $10b contracts - The Washington Post

Most overvalued stock in history, pushing $400b EV, $10b in earnings are not really in sight.

Does Apple lack strategic vision, and is time is running out for Tim Cook? - UnHerd

Is Apple building its own AI ‘answer engine’? - TechCrunch

Anthropic's 'vaccine of evil', expose your model to bad stuff in training to get a clear fingerprint of undesirable intrusive thoughts, then dial back the evil personality vector - Business Insider

Bill Gates warns AI automation cut entry-level jobs by 35% since 2023, doesn't matter how well the kids know AI - Fortune

The AI job cuts are accelerating - Financial Times

Job-seekers increasingly refuse AI-only interviews, calling them impersonal despite HR claims of efficiency gains. - Fortune

Who is the guy who rejected $1.5b from Zuck? - The Times of India

then there is the guy who turned down $125m but accepted when Zuck doubled it - The Times of India

North Korean remote IT workers infiltrated 320 companies using GenAI - Fortune

With the right tools, somebody who barely speaks English can take your job

San Francisco proper overtakes suburban Silicon Valley for AI startups - The New York Times

Silicon Valley in hard tech era, no more Web 2.0 social network crap - The New York Times

The latest crop of early-20s dropout startup CEOs - The New York Times

AI LLMs are now so clever that they can independently plan and execute cyberattacks without human intervention and I fear that it is only going to get worse - TechRadar

Security folks created an agent red teaming data set (good), write clickbait paper (bad). - THE DECODER

Any LLM-based guardrail prompts that say 'don't do X' can be breached. You must block bad stuff via access controls. Untrusted prompt + data access + tools with ability to do bad stuff = bad time

AI Tools Gave False Information About Tsunami Advisories - SFGate

Drones and AI found the body of a hiker who had been missing for a year - WIRED

AI, Satellites and drone swarms: The new high-tech quest to fight wildfires - MSN

Predicting the path of malaria with machine learning - Nature

UCSF study finds AI-generated hospital discharge summaries are short, clear but can miss important details - Medscape

AI obits are the biggest new trend in "death care." Tens of thousands of AI obits have been made already. Often the families don't even know - The Washington Post

Friends, bots, AIs, lend me your ears

Should conversations with AI be confidential? - ctech

How private are your chats? here's who can see them - Fast Company

ChatGPT Study Mode excels at interactive tutoring while Google’s NotebookLM offers strong source-grounded summaries and unique audio discussions. - Tom's Guide

Agents are great but garbage in, garbage out, a lot of companies lack data quality. The good news is, AI can help quite a bit with finding and addressing data inconsistencies. - TechRadar

By training neural networks on distorted light patterns from objects behind turbid media, an AI system can estimate object positions with precision nearly matching the theoretical maximum defined by Fisher information, a measure of the ultimate achievable accuracy. - SciTech Daily

A 2b-artificial-neuron neuromorphic computer - South China Morning Post

Hierarchical reasoning models beat LLMs on some tasks - TechRepublic

If you always wanted to produce your own TV show, now is your big chance - Geeky Gadgets

Artificial intelligence is now deeply integrated into the film industry, assisting in tasks from de-aging actors and generating special effects in blockbusters to enabling independent filmmakers to create entire films with minimal resources using AI tools like Runway and Kaiber AI. - Domusweb

Disney experimented with AI technology to create a digital double of Dwayne Johnson for the live-action Moana using deepfake technology but ultimately decided not to use the footage due to legal, copyright, and security concerns after 18 months of negotiations with AI company Metaphysic. - MSN

OpenTable Launches Embedded Concierge to Answer Diner Questions and Drive Restaurant Bookings - Restaurant Tech News

Airlines use AI for price discrimination, prompting transparency and fairness calls; should consumers use cash, VPNs, and clear browser data? - Fast Company

Your next direct report might not be human. Are you ready to lead it? - Fast Company

“I managed to pass the course… but I realized that I couldn’t remember the last time I had written an essay by myself, which was my favorite activity. This was the catalyst for me to stop using the app” - EL PAÍS English

langextract , write some prompts and it extracts and highlights the sections that match in text. - GitHub

Follow the latest AI headlines via SkynetAndChill.com on Bluesky