How Are Developers Using AI? Inside the 2025 DORA Report (Category AI)
ChatGPT Pulse, which makes recommendations based on my interests, pointed me to this article from 23 September. I have discussed DORA research and AI’s effect on software development with ChatGPT quite a lot. At T-Bank we have even launched a large study of AI in Russia’s engineering culture. The 2025 DORA report is now available, along with an executive summary in this Google blog article. Here is a brief account of the survey findings.
1. AI use is almost universal
- Around 90% of developers use AI tools.
- More than 80% believe they improve productivity.
- Around 59% believe they improve code quality.
A majority, 65%, rely substantially on AI in software development: 37% report a moderate degree of reliance, 20% a large degree and 8% a very large degree.
2. The trust–productivity paradox Trust is more limited than adoption:
- 24% report a great deal (4%) or a lot (20%) of trust in AI.
- The Russian note lists 30% for limited trust, with 23% specifically described as trusting AI “a little.”
People can find AI useful without fully trusting it. They may have organised their work so that it acts as an assistant rather than taking complete responsibility for a task.
3. Team archetypes This year’s research describes AI as a “mirror and multiplier.” It improves effectiveness in cohesive organisations and exposes weaknesses in fragmented ones. To explore this, the authors identify seven team archetypes, ranging from harmonious, successful teams to those facing legacy bottlenecks.
4. A blueprint for guiding AI in organisations The report proposes the DORA AI Capabilities Model, identifying seven foundational practices for effective AI use.
#AI #Software #Engineering #Management