Sam Blechman, BS

"Imagine you’re admitted to the ER with a fever and low blood pressure. The hospital’s fancy new AI model flags you as having a 70% chance of sepsis. Sounds bad, right? But what if that 70% chance doesn’t mean what the providers think it means? What if, historically, only half of the people the model called “70% likely” actually had sepsis? That’s what we mean when we say a model is miscalibrated—its predicted probabilities don’t line up with reality."

"A set of fundamental concepts in statistics and machine learning can be traced back to a deceptively simple 2x2 table: the confusion matrix. Despite its simplicity, the confusion matrix is a powerful tool that appears in two key domains of statistics: binary classification and hypothesis testing. In both cases, it serves a similar function: it measures how often and in what direction we are incorrect."

"Developed by researchers at Google DeepMind and Isomorphic Labs, AlphaFold3 improved upon its predecessors by modeling protein-ligand interactions at far greater accuracy than previous technologies, among other things. But instead of being met with universal acclaim, the article left a bad taste in the mouths of many within the scientific community."