# DORA 2025: AI Is an Amplifier, Not a Silver Bullet
Key findings from Google's DORA 2025 report: 90% of software professionals are using AI, 59% report improved code quality — but AI amplifies both good and bad engineering practices.
Key Data Points
- 90% of software professionals use AI tools (2025 DORA Report)
- 59% report improved code quality
- AI is an amplifier: good processes get faster, bad processes get messier
Adapting the Four-Dimension Framework for AI
| DORA Metric | How It Changes in the AI4SE Era |
|---|---|
| Deployment frequency | AI accelerates coding, but review becomes the new bottleneck |
| Change lead time | Coding time ↓, but review + testing time share ↑ |
| Change failure rate | AI-generated code’s failure rate is on par with or slightly below human code |
| Recovery time (MTTR) | AI can assist diagnosis, but fix decisions still require humans |
The Need for New Metrics
The DORA four dimensions aren’t enough anymore — AI4SE needs to add:
- Agent throughput: number of effective tasks completed per agent per day
- Spec pass rate: proportion of agent-generated code that passes spec acceptance on the first try
- Human intervention rate: proportion of agent work requiring human intervention
- Independent review finding rate: proportion of issues found by an independent evaluator
References
- Google DORA 2025: State of AI-assisted Software Development
- DORA AI Capabilities Model
- SPACE Framework vs. DORA comparison