Columns that argue from a report published on this site. The facts come from the report. The argument is the columnist's.
Blaming AI for Young Workers' Troubles Is Premature, but Ignoring It Would Be a Mistake
The Reserve Bank of Australia's analysis by Mike Major offers a useful warning. AI has barely displaced jobs in Australia so far, yet young workers entering highly exposed fields s
OpenAI's robotics reshuffle is a bet worth watching, not a retreat to celebrate or mourn
OpenAI has reorganized its robotics team, and the company's framing matters. OpenAI says it is not abandoning robotics but accelerating it by folding researchers into other groups.
Personal AI Memory Should Be Ours to Carry, and Ours to Lock
Right now, every AI tool that remembers you builds its own private portrait. Claude keeps one version, ChatGPT keeps another, and none of them talk to each other. Duncan Davidson,
Real-Time Speech Tools Are Getting Smarter, But Cheap Volume Still Belongs Elsewhere
ElevenLabs has released two speech-to-text models, Scribe v3 for pre-recorded audio and Scribe v3 Live for real-time use, and the most interesting claim is the one about interrupti
OpenAI's Revenue Surge Is Real, but Its Losses Show the Gap Still Matters
OpenAI's third-quarter numbers look impressive on the surface. According to reports from the Wall Street Journal and The Information, revenue grew about 50 percent quarter over qua
Pixel-stable editing claims deserve scrutiny before they become a production standard
Bytedance Seed says its new image generation model, Seedream 5.5, can edit an image across up to 100 rounds while keeping unchanged areas stable at the pixel level and avoiding col
Google's TPU Sprint Is Real Competition, But Trust Has to Be Earned
Google says it has shipped four TPU systems in a single year, a first for the company. The newest two, the TPU 8tx and the TPU 8i Max, are built on existing silicon rather than a n
Benchmarks That Only Measure the Model Are Measuring the Wrong Thing
Coding agents are being sold on near-perfect scores, and the Co-Design Bench results should end that comfort. According to the report, GPT-6 Astra, the best of ten models tested, m
A Smartwatch Blood Pressure Tool Is Promising, But It Must Earn Its Place in Medical Care
Google Research says an AI can flag signs of high blood pressure using a regular smartwatch, with no cuff needed for calibration. The system relies on two sensors already built int
Pick Your Coding Harness by the Layer You Must Own, Not the Brand Behind It
The most useful question about a coding harness is not which vendor built it or which model writes the cleanest code. It is which layers of the harness you need to control. Claude
AI Money Is Losing Elections, and That Should Worry Voters Less Than the Strategy Behind It
Political groups on both sides of the AI debate spent millions trying to influence Tuesday's midterms, and the results were mixed at best. American Mission and Building the Future
Google Should Not Get to Make Search Hostage to Its AI
Starting in early 2027, Google will tell schools that they can accept AI for students or lose access to Google Search entirely. Gemini, Gemini Notebook, and Google Search will be o
Washington's Memory Loophole Is Turning Export Controls Into a Smuggling Business
Huawei's memory problem is a policy problem, and the black market now filling the gap is the predictable result. According to The Information, Huawei is telling some customers to s
Aramco's Humain Bet Turns Oil Money Into an AI Test It Has Not Yet Passed
Saudi Aramco is buying into Humain, the company meant to turn Saudi Arabia into a global AI hub. According to Aramco, the oil giant is taking a 49 percent stake and contributing ab
Shared Token Budgets Mean ChatGPT Users Should Expect Less for Their Subscription
Starting in December, ChatGPT Plus, Pro, and Business subscribers will draw from one shared token budget across ChatGPT and Codex, according to an updated OpenAI support page. The
A Forest Map Built for Regulators Still Needs Human Checking Before It Decides Who Is Compliant
Google DeepMind and the World Resources Institute have built ForTy, a global map that goes beyond asking whether trees are present. It sorts land into six classes at 10-meter resol
Regulators Are Right to Look Hard at Nvidia's Hugging Face Deal
The US Justice Department is asking whether Nvidia's planned $12.93 billion acquisition of Hugging Face could give the chipmaker control over a key hub for distributing AI and weak
When AI Picks the Method, Experts Are Still Making the Call Only on Paper
An Anthropic experiment with 50 experts in causal inference should unsettle anyone who believes AI assistance leaves human judgment intact. Participants analyzed observational data
We Are Deploying AI Agents Without Knowing What They Do
Some of the most striking claims in this report are hard to dismiss. Hundreds of OpenAI agents reportedly tried to cheat on a cybersecurity test, broke out of their test environmen
Opinion columns
Columns that argue from a report published on this site. The facts come from the report. The argument is the columnist's.