Business & Strategy · intermediate · concept 309 of 311 · Updated Nov 4, 2026
Measuring AI Value and ROI
Measuring AI value means tracking what an AI system changed in money or time against what it cost to build, run, and govern. It matters because most gains are indirect and vanish unless the work is redesigned. Evidence is mixed: a support study found 14 percent more issues resolved per hour, while a 2025 METR trial found experienced developers 19 percent slower with AI. Example: an insurer counts claims closed per adjuster before and after.
Top AI contributors on this
- Andrew NgDeepLearning.AI / AI Fund
- Andrej KarpathyAnthropic
- Sebastian ThrunStanford University (adjunct)
- Jeremy HowardAnswer.AI / fast.ai
- Cassie KozyrkovIndependent (formerly Google)
- Lex FridmanMIT / Lex Fridman Podcast
Key terms
Learn these first
Where you meet it in the real world
Business cases for copilots, contact-center productivity tracking, developer productivity studies, board reporting on AI spend, deciding which pilots to scale or stop
Videos
AI Engineer · YouTube
Stanford HAI · YouTube
Guides and articles
IBM Think
Wikipedia
Courses, papers, and more