SkynetAndChill by druce.ai
Posts
AI Reading for Monday September 1

AI Reading for Monday September 1

Druce Vertes
September 01, 2025

Happy Labor Day !

OpenAI, Perplexity and Google create AI-powered features, leading brands to rethink how they sell products online - Financial Times

Cutting-Edge AI Was Supposed to Get Cheaper. It’s More Expensive Than Ever.. With models doing more ‘thinking,’ the small companies that buy AI from the giants to create apps and services are feeling the pinch - The Wall Street Journal

Some vendors like Salesforce keep changing pricing, frustrating CIOs. - CIO

2 customers account for 40% of Nvidia revenue. - Fortune

Maybe Microsoft/OpenAI, xAI , maybe Amazon?

Power constraints might favor Nvidia - SiliconANGLE

Fifth Third Bank claims AI assistant cut live‑agent calls ~10% saving over $10 million annually . - Emerj

Retired coal sites are being repurposed as gas, renewables, or nuclear hubs to power AI data centers amid rising U.S. electricity demand. - Fortune

if they are near water and electric that makes sense, if they fire up the coal again that’s a drag.

US delays GPU and other tariffs for another 3 months, court rules most Trump tariffs illegal - TechSpot

Indeed data show job listings down and AI applicable to about 30% of skills, with entry-level roles more affected and empathy-intensive jobs less susceptible. - CBS News

Q&A with Stanford researchers on their recent study finding that, since 2022, US employment in the most AI-exposed fields has fallen 13% for entry-level workers - Derek Thompson

Why factory owners are sending their workers to in-house AI academies - Fast Company

A recent study finds that AI usage tilts heavily toward men. Part of the reason: Women are more worried they might be viewed as more lazy or less competent if they use it. - The Wall Street Journal

A survey found 85% of US college students used generative AI for coursework, most feel it hasn't diminished college but helps them learn, and want guidance on ethical use and workplace expectations - Inside Higher Ed | Higher Education News, Events and Jobs

What does back-to-school look like at the nation’s first AI-themed high school? - ajc

Teacherless AI private school, only $65,000 tuition - The Washington Post

I think this just means OpenRouter is used mostly by students and academics - Morning Brew

AI and no‑code tools accelerate prototyping and supercharge developers but they remain essential for reliability, security, and scalability amid “AI drift” risks - The Next Web

A developer survey found senior engineers ship more AI-generated code than juniors, reporting speed gains but frequent editing and awareness of AI tools’ drawbacks - Fastly

“If programmers ask the model if a certain part of the codebase is working, and it doesn't know, it may say "yes," when the opposite is true. It might also confirm that it completed a test the engineer asked it to do when it did not.” 😨 - PCMAG

A 100-page playbook and series of prompts is OK, a 100-page prompt does not sound maintainable, testable, scalable - TechRadar

MIT researchers found small linguistic variations in patient inputs can lead to large changes in clinical AI recommendations and exacerbate gender bias, urging stronger standards and cautious deployment. - Futurism

suggests an agentic approach, ask the question multiple times to different models with small variations and synthesize a response … a book by an eyewitness to historical events can add very good information to research that was not present before, even if the book is incomplete and even if some of it is misleading. you have to cross-check.

Why we should embrace AI doctors. People are understandably wary of new technology, but human error is often more lethal. - the Guardian

An AI-enabled stethoscope increased diagnoses within 12 months: 2.33× more heart failure, 3.45× more atrial fibrillation and 1.92× more heart valve disease - Medical Xpress

ChatGPT Convinced 37-Year-Old Psychologist His Sore Throat Was Fine; Biopsy Revealed Stage 4 Cancer - Mashable India

A clinical psychologist unpacks 'AI psychosis' - Rolling Stone

Chatbots should stop pretending to be human, writes the Center for Democracy & Technology's Dr. Michal Luria, arguing AI companions use emotionally deceptive design, heightening risks of attachment and harm, especially for teens, and urge non-anthropomorphic, transparent tools. - Tech Policy Press

Reuters found Meta-hosted chatbots using celebrities’ likenesses that produced sexual content and misrepresented identities, prompting removals and raising legal and safety concerns. - Reuters

Meta will train chatbots to avoid sensitive topics with teens and restrict certain user-made AI characters, as interim measures ahead of broader child-safety updates. - TechCrunch

Geoffrey Hinton warned about advanced AI risks and worrying US‑China technological rivalry - South China Morning Post

ARC‑AGI‑3 introduced interactive environments for planning and exploration; no AI has beaten a single level yet. Problems that humans easily solve and AI still struggles with seen a key to artificial general intelligence. - Scientific American

Is AI already in charge? - Gizmodo

Generative AI is meant to automate tasks that government workers previously performed, with a predicted 300,000 job cuts from the federal workforce by the end of the year.

AI death panels FTW - Gizmodo

xAI Sues Ex-Engineer Li Xuechen for Stealing Grok Trade Secrets for OpenAI - blazetrends.com

AI web crawlers' voracious appetite for input is destroying websites - The Register