
Throughput gains vs. stability risks

Measuring the impact of AI on software engineering – with Laura Tacho
Topic 10 of 14
Published Jul 23, 2025 · Clip 52:03–55:15 (3:12)
Laura explains DX's core measurement framework and how AI acts as an enabler whose impact shows up in existing fundamentals like speed rather than requiring entirely new metrics. DX data shows diffs per engineer increase in both volume and complexity—AI users tackle more complex work. The host predicts bugs will increase with larger diffs, and Laura confirms with Dora's AI impact study: delivery throughput is already slowing slightly, likely due to increasing batch size, and a 25% increase in AI adoption is forecasted to bring a 7.2% reduction in delivery stability. She stresses that tunnel vision on lines of code or acceptance rate misses quality, stability, and maintainability.