Nearly 90% of technology professionals now use artificial intelligence in their work. But according to the 2025 DORA State of AI-assisted Software Development report, there’s still a significant gap in trust between developers and the tools they increasingly rely upon. The report surveyed nearly 5,000 technology professionals globally and found that while AI adoption has become “nearly universal,” there are still some fundamental organisational challenges.
The research suggests that AI amplifies the strengths or weaknesses in organisations, depending on whether they are high-performing or not. The report finds that “AI’s primary role in software development is that of an amplifier. It magnifies the strengths of high-performing organisations and the dysfunctions of struggling ones.” It also mentions a widespread assumption that using AI tools can drive organisational transformation, which this finding dispels.
The 2025 Developer Survey conducted by Stack Overflow shows a growing distrust among developers, which supports these worries. The survey results show developers distrust AI tool accuracy at 46% while trust levels reach only 33%. Only 3% of developers report “high trust” in AI-generated output, suggesting widespread scepticism about the quality of AI-assisted code despite its prevalent use.
Commenting on the DORA report in a LinkedIn post, Dr Laura Weis observes that “faster doesn’t always mean better. The 2025 DORA report shows it clearly: AI helps people push out more work. But the real headaches – burnout, broken processes, clunky cultures – don’t go away. In some teams, the pressure just ramps up: more output expected, same resources, same stress.”
The DORA research shows that software stability has become a significant issue since generative AI started to be adopted. The current positive relationship between AI adoption and delivery throughput marks a change from previous results, yet software delivery instability continues to rise. The research indicates that teams have adjusted their development speed, but their systems lack the necessary capabilities to handle AI-driven development safely.
The research team investigated whether AI speed improvements would compensate for instability through rapid failure and quick repair methods, but their analysis showed no positive results. The researchers believed that the additional speed from AI would help counterbalance the disorder by using the ‘fail fast, fix fast’ concept. However, the report found that this approach does not deliver the intended benefits. Dr Weis explains that instability continues to damage product quality and employee performance.
The research shows that AI implementation does not affect workplace tension or developer exhaustion levels, but developers achieve higher productivity results by using it. The ongoing nature of these problems indicates that they stem from existing organisational systems and can’t easily be solved by implementing individual productivity tools.
The DORA team created an AI Capabilities Model to analyse these systemic problems through seven essential organisational practices that amplify AI’s benefits. The seven capabilities of the model operate at team and organisational levels instead of focusing on individual tool usage. They require organisations to establish clear AI policies and maintain healthy data ecosystems and quality internal platforms.
The research reveals that organisations with a user-centric focus see amplified benefits from AI adoption, whilst those lacking this focus often see negative impacts on their teams’ performance.
Platform engineering has emerged as a critical foundation, with 90% of organisations now having adopted internal platforms and 76% maintaining dedicated platform teams. The research shows that high-quality platforms are essential enablers for scaling AI benefits across organisations, providing necessary guardrails and shared capabilities.
However, the transition is not without trade-offs. High-quality platforms correlate with slight increases in software delivery instability, which researchers interpret as “risk compensation” where organisations with strong recovery capabilities can afford to experiment more whilst maintaining overall system reliability.
The report establishes seven team performance profiles to analyse the intricate relationships between performance, stability, and team well-being. As explained by DORA lead Nathen Harvey in the report, the seven team performance profiles span from “harmonious high-achievers” who excel in all areas to teams dealing with “foundational challenges” that show major deficiencies in their processes and results.
Leah Brown from IT Revolution analyses the findings and emphasises that “the research demolishes the notion that AI adoption is simply a tools problem. Instead, it reveals AI success as fundamentally a systems problem requiring organisational transformation.”
Value stream management emerges as a critical practice for maximising AI investments. The research shows that organisations with mature value stream management practices see dramatically amplified benefits from AI adoption on organisational performance, helping ensure that individual improvements translate into broader organisational advantages.
The findings align with established patterns from previous technology transformations. Brown notes that “organisations that simply moved to cloud infrastructure without rethinking architecture saw limited returns, while those that restructured their applications, teams, and operations unlocked real value.”
However, individual developers are reporting many benefits to using AI despite these challenges. More than 80% of survey respondents think that AI has increased their productivity, while 59% observe positive impacts on code quality. The most common AI use case remains writing new code, with 71% of code writers using AI assistance.
Dr Weis concludes that the pattern remains consistent:
healthy teams climb higher, shaky ones fall faster. Tools don’t change that. They just shine a light on it.
– Dr Laura Weis
The research suggests that organisations must treat AI adoption as a comprehensive transformation effort rather than a simple tool deployment. Success requires investing in foundational systems, including platforms, data ecosystems, and engineering disciplines, that can amplify AI’s benefits while addressing the organisational factors that currently limit them.
The full comprehensive 142-page report is available to download from DORA.
