Fastly survey: Senior developers report shipping more AI‑generated code than juniors
Fastly’s July 2025 Developer Survey reports differences in how AI coding tools are used across experience levels. Among 791 professional developers surveyed, 32% of senior engineers (10+ years of experience) say over half their shipped code is AI‑generated, compared to 13% of junior developers (0–2 years). The results indicate that more experienced engineers report greater reliance on AI and higher rates of AI‑assisted code in production.
Fastly published the results in a summary of its July 2025 poll of developers in the US, with responses collected July 10–14. All respondents reported that writing or reviewing code is a core part of their job. As with all self‑reported data, some bias is possible. The full post is available at Fastly: https://www.fastly.com/blog/senior-developers-ship-more-ai-code?utm_source=tldrnewsletter
Usage patterns split by experience
Senior developers were more likely to report investing time in fixing AI output, while still reporting net speed gains. Just under 30% of seniors said they edit AI‑generated code enough to offset most time savings, compared to 17% of juniors. Despite the additional editing, 59% of seniors report that AI tools help them ship faster overall, versus 49% of juniors.
Qualitative responses describe benefits and trade‑offs. “AI will bench test code and find errors much faster than a human, repairing them seamlessly. This has been the case many times,” a senior developer said. A junior respondent described challenges: “It’s always hard when AI assumes what I’m doing and that’s not the case, so I have to go back and redo it myself.”
Speed gains: moderate vs. significant
Reported speed gains vary by seniority. Just over half of juniors say AI makes them moderately faster, while only 39% of seniors say the same. However, 26% of seniors say AI makes them a lot faster, roughly double the 13% of juniors who agree. The report suggests that experience may equip senior engineers to spot subtle AI mistakes and course‑correct quickly, enabling more confident use in production and on higher‑stakes work.
This aligns with what reaches production: 32% of seniors say more than half of their shipped code is AI‑generated, versus 13% among juniors—a gap that persists despite concerns about “vibe coding” and potential vulnerabilities.
Perception vs. reality on productivity
Across all respondents, 28% say they frequently have to fix or edit AI‑generated code to the point it offsets most of the time savings, while only 14% say they rarely need to make changes. Even so, more than half still report being faster with tools such as Copilot, Gemini, or Claude.
External research cited by Fastly adds a counterpoint. A recent randomized controlled trial of experienced open‑source developers found that developers using AI tools took 19% longer to complete tasks, despite the perceived momentum from autocompletes: https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/
Developer comments describe the gap between initial speed and downstream rework. “An AI coding tool like GitHub Copilot greatly helps my workflow by suggesting code snippets and even entire functions. However, it once generated a complex algorithm that seemed correct but contained a subtle bug, leading to several hours of debugging,” one respondent noted. Another added: “The AI tool saves time by using boilerplate code, but it also needs manual fixes for inefficiencies, which keep productivity in check.”
Job satisfaction trends
Despite mixed efficiency outcomes, nearly 80% of developers say AI tools make coding more enjoyable. Respondents cited easing roadblocks and reducing boilerplate: “It helps me complete a task that I’m stuck with. It allows me to find the answers necessary to finish the task,” one survey participant said. Enjoyment does not necessarily equate to throughput, but the data indicates increased reported enjoyment.
Sustainability and AI awareness
The survey also examined sustainability practices. Green coding considerations increase with experience: 56% of juniors say they actively consider energy use in their work, rising to nearly 80% among mid‑ and senior‑level engineers. Awareness of AI’s environmental cost appears broad, with roughly two‑thirds across experience levels acknowledging that AI tools can carry a significant carbon footprint. Only a small minority—under 8% even at the most junior levels—reported being completely unaware.
Methodology
Fastly conducted the survey from July 10 to July 14, 2025, collecting responses from 791 professional developers in the US. All respondents confirmed that writing or reviewing code is a core part of their job. The company notes quality controls were applied; however, as with any self‑reported survey, some bias is possible.