"AI does not automatically improve software delivery performance. Instead, it acts as a multiplier of existing engineering conditions, strengthening high-performing teams while exposing weaknesses in organizations with fragmented processes and poorly structured development systems."
"The success of AI in software engineering depends less on the sophistication of the tools and more on the strength of the organizational systems surrounding them. Engineering culture, platform capabilities, development workflows, and internal knowledge systems ultimately determine whether AI improves productivity and delivery outcomes or simply accelerates complexity."
"Approximately ninety percent of developers now report using some form of AI assistance in their work. Around two-thirds say they rely heavily on these tools for tasks such as writing code, generating documentation, debugging problems, or exploring unfamiliar frameworks."
"While developers frequently rely on AI to accelerate development tasks, a significant portion remain cautious about the accuracy and reliability of AI-generated code. This hesitation reflects broader concerns around maintainability, correctness, and long-term sustainability."
The 2025 DORA report examines AI's impact on software development through nearly 5,000 survey responses and over 100 hours of interviews. While 90% of developers use AI assistance and report productivity gains, AI does not automatically improve delivery performance. Instead, it amplifies existing conditions—strengthening high-performing teams while exposing weaknesses in organizations with fragmented processes. Success depends on engineering culture, platform capabilities, workflows, and knowledge systems rather than tool sophistication. Developers show measurable improvements in problem-solving speed and code efficiency, yet significant hesitation remains regarding AI-generated code accuracy and reliability, creating tension between productivity gains and trust concerns.
#ai-assisted-software-development #organizational-systems-and-culture #developer-productivity #code-quality-and-trust #devops-performance
Read at InfoQ
Unable to calculate read time
Collection
[
|
...
]