“The Data That Knows You Better Than You Do”
👉 For more insights, check out the full book and audio summary here 📖
📕 Transcript and audio of this episode is available here for free
Title: Predictive Analytics
Author: Eric Siegel
Category: Sales, Branding & Marketing, AI, Data & the Future
What if a computer could figure out you were pregnant before your own father did? Sam and Sophie open with the Target story that made everyone uneasy for a reason.
They walk through Eric Siegel's core idea — past behavior predicts future behavior — and the machine learning engine underneath it: labeled historical examples, pattern-finding at massive scale, and probability scores that decide who gets a catalog, a coupon, or a loan. The conversation turns on the Target pregnancy model, the healthcare version where risk-flagging clearly helps, and the criminal justice feedback loop where arrest data trains models that send more police to the same neighborhoods. Siegel's answer isn't to throw the tech out; it's transparency, accountability, and understanding that models find correlations, not causes.
If you've ever wondered why an ad seemed to read your mind, this episode names the machinery and the tradeoffs. The takeaway: accuracy and fairness are separate questions, and a score is just the past laundered into a forecast.
Predictive Analytics by Eric Siegel. For the full written summary, head to 7minutebooks.com/app — over 6,000 fiction and nonfiction titles you can read or listen to in any language, with infographics on many titles. Unlimited access from $2.99 a month, $9.99 a year, or $19.99 lifetime.
Chapters
00:00Past Behavior Predicts Future Behavior02:52Target's Pregnancy Prediction Model03:55Healthcare, Bail, And Feedback Loops05:24Correlation Is Not Causation06:38Overfitting And The Accuracy Tradeoff07:30Invisible Scoring And The Human Question
Topics
Related episodes
