What Measurements Don't Show
A number comes into being under certain conditions. This series asks what is left of it when those conditions fall away.
- Benchmark Performance ≠ Bedside PerformanceAn AI model diagnoses with 87% accuracy, and it can still be dangerous in the clinic. Why study numbers don’t tell us what we want to hear, and what counts instead.From Science · 5 min read
- Validation Is No Longer EnoughA scalpel sits on a shelf until someone needs it. An AI model keeps learning, drifts, adapts, and in doing so falls through every grid built for static medical devices. Why the EU approval system has to be fundamentally rethought for adaptive AI.From Science · 5 min read
- 37 Percent of AI Citations Don't ExistOscar works in peer review at a medical journal. What he now sees there: submissions whose references lead nowhere, author names that are wrong, titles that were pulled out of thin air. A survey of a silent crisis.From Science · 4 min read
- The Demo and the TuesdayAlmost every demo of an AI system shows the same cutout: the successful run. That isn’t fraud, it’s a format. You just should know what it leaves out.On Stage · 3 min read