Human impact
Health, creativity, productivity: what AI concretely changes in people's lives and work.
Health · Creativity · Productivity — 36 articles
29 English hospitals "benefit" from Palantir. They say they don't use it
It all depends on what counts as a benefit.
Alexandre Noto · Nov 3, 2026 · 6 min
Human impactExperts follow an AI agent whose advice is a coin toss
The funniest part is how they justified it.
Julien-Pierre Noto · Oct 30, 2026 · 6 min
Human impactMicrosoft's AI diagnoses twice as well as doctors?
In 2025, the same study said four times.
Alexandre Noto · Oct 23, 2026 · 6 min
Human impactPostgreSQL 19 will ship without about ten of its new features
The code was written without AI. It was reviewed with it.
Alexandre Noto · Oct 17, 2026 · 6 min
Human impactOne in ten Linux kernel patches says an AI helped
Even people who've been writing Linux code for twenty years have joined in.
Alexandre Noto · Oct 16, 2026 · 6 min
Human impactA developer cloned Photoshop with Claude. Is that legal?
We opened the repo, and Adobe's terms of use.
Alexandre Noto · Oct 13, 2026 · 6 min
Human impactNine detectors analyzed the same pages of Orélien's novel
How many of them actually said 100% AI?
Katja Liersch · Oct 1, 2026 · 10 min
The Verified NumberAI Makes a Genetics Discovery in Record Time
A breakthrough, or a lucky break?
Alexandre Noto · Sep 29, 2026 · 6 min
Human impactJev, the AI built to make decisions
Why Jev has developers buzzing
Alexandre Noto · Sep 21, 2026 · 7 min
Human impactAn AI Wrote a Diagnosis This Patient Never Had
Who's checking what the AI writes in your medical record?
Katja Liersch · Aug 31, 2026 · 6 min
The Verified Number45% of Workers Skip AI Checks? The Report Says 60%
The report and the press release come from the same firm. On the exact same question, they don't show the same number, and only one of the two made it into the news in France.
Alexandre Noto · Aug 18, 2026 · 6 min
The Verified NumberWhat the Blind Test on AI-Written Fiction Actually Shows
1.54 versus 0.97, on a scale that runs from minus three to plus three. Both scores are positive, and nobody could reliably tell which story came from a machine.
Alexandre Noto · Aug 11, 2026 · 6 min
Human impactAI Doesn't Make You Dumber. It Makes You Dependent.
A Brown professor's students averaged 96% with AI. Stripped of it, they collapsed to 48%. That gap doesn't measure how smart they are. It measures what's left once the tool gets unplugged.
Katja Liersch · Jul 12, 2026 · 6 min
Human impactAI is great... at first
AI assistants shine in the first sessions, then fall off once a project drags on. Here's why, and what to do about it.
Julien-Pierre Noto · Jul 8, 2026 · 6 min
Human impactAI lifts homework grades, sinks exam scores
A study of 26,811 Chinese students measures what AI does to learning: homework grades climb, exam scores fall.
Katja Liersch · Jul 4, 2026 · 5 min
Human impactAt 350 words, AI passes for a novelist
A blind test pitted GPT-5 against four award-winning novelists. Readers couldn't tell them apart. Here's what still gives the machine away.
Katja Liersch · Jul 4, 2026 · 5 min
Human impactMedical AI: What Is a Specialized Model Worth?
A study pits FDA-cleared medical AI against consumer chatbots. The verdict raises an uncomfortable question about what the label actually buys.
Katja Liersch · Jun 25, 2026 · 5 min
Human impactPublishing Doesn't Fear AI. It Fears Volume.
The real threat AI poses to publishing isn't the quality of what it writes, but the unlimited quantity it produces for next to nothing.
Katja Liersch · Jun 22, 2026 · 5 min
Human impactAdobe automates the plumbing, not the craft
Adobe is rolling out AI agents in Photoshop and Premiere. A look at what they actually do, and who profits first.
Katja Liersch · Jun 18, 2026 · 5 min
Human impactThe real price of AI never shows up in the demo
Emissions underestimated by a factor of 500, surprise cloud bills hitting $127,000: the costs AI demos always leave offstage.
Julien-Pierre Noto · May 18, 2026 · 5 min
Human impactOntario AI Scribes: 60% Recorded the Wrong Medication
Ontario's Auditor General tested 20 AI scribes approved by the province. Average accuracy: 12 out of 20. The easiest use case for AI in healthcare just collapsed at scale.
Katja Liersch · May 17, 2026 · 5 min
Human impact28 days to a perfect body: how AI fake fitness coaches scale
AI-generated fitness coaches flood TikTok and Instagram with impossible promises. The real damage is not the scam, it's what it does to teen mental health.
Katja Liersch · May 8, 2026 · 5 min
Human impact600,000 Unanswered Questions: How AI Is Filling (and Exploiting) Medical Deserts
On the same day: ChatGPT handles 600,000 medical queries per week from hospital deserts, and a two-person startup made $1.8 billion with fake AI doctors. Not a coincidence: the same void, two very different uses.
Katja Liersch · Apr 7, 2026 · 4 min
Human impact3 principles to use ChatGPT better than 90% of people
You use ChatGPT every day and the results keep disappointing you. Odds are, the problem isn't the AI. Here are the 3 principles that change everything.
Julien-Pierre Noto · Mar 26, 2026 · 5 min