The surprising ways AI helps strong dev teams and hurts weak ones, according to Google (www.zdnet.com)

Google’s 2025 DORA report (5,000 developer survey responses plus 100+ hours of interviews) finds AI has moved from hype to mainstream: 90–95% of developers use AI tools, median interaction is ~2 hours/day, 60% use AI “about half the time” or more, and 80% report higher productivity though only 59% see improved code quality. Trust in AI sits at ~70%. Notably, DORA reverses last year’s finding (where AI reduced productivity) and identifies AI as an “amplifier” — it magnifies existing strengths and dysfunctions in teams rather than acting as a neutral productivity booster. The report’s actionable takeaway for AI/ML teams is systems, not just models: successful AI adoption requires platform engineering and value-stream management (VSM). Organizations with strong internal platforms and VSM practices convert localized AI gains into organization-wide improvements; weak platforms often lead to more downstream chaos. DORA also highlights seven enabling practices — clear AI policy, high-quality/accessible data ecosystems, version control, small-batch work, user focus, and shared quality platforms — as prerequisites for reliable AI-driven delivery. For practitioners, the implication is clear: invest in data hygiene, platform APIs, VSM tooling, and engineering disciplines (CI/CD, tests, revision control) to harness AI safely; without those foundations, AI risks amplifying instability.