In today’s world of random forests, deep learning, and ensemble models, we may not use AIC directly—but the thinking behind it, and its connection to CV, still shapes how we assess models.

📚 A 47-year-old result still quietly powering modern ML workflows.