
Measuring the impact of learning and AI

From Knowledge to Wisdom 🧠 — with Hywel Carver
Topic 6 of 7
Published Nov 7, 2025 · Clip 40:01–53:07 (13:05)
Hywel argues learning should always have a measure tied to the business outcome it serves — e.g., p99 latency dropping 10%, or a client whose cycle time fell to 40% and velocity increased ~230% after API-first training. He criticizes login-rate metrics (73% login vs ~13% average e-learning completion as meaningless). He then introduces a study with DX measuring the holistic impact of Skiller Whale's generative AI curriculum on productivity, developer satisfaction, flow, and code churn, discussing how AI helps around meetings but disrupts flow, and why '90% of code written by AI' is a misleading metric.