Senior developers use AI more extensively for coding tasks compared to junior developers
The data is clear: developers with more than 10 years of experience are leaning heavily into AI. Nearly one-third of senior developers now report that over half of their shipped code is generated by AI. That’s a massive signal. These developers aren’t experimenting; they’re relying on AI to build production code at scale.
Why does this matter? Because it shows that the people most capable of evaluating the integrity of AI-generated output are not just tolerating it, they’re trusting it. Senior engineers have enough experience to spot weaknesses. If they’re still choosing to let AI drive a significant amount of the codebase, it tells us that the performance of these tools is operational.
At the same time, this kind of large-scale adoption exposes a crucial point for leadership: success with AI depends on talent maturity. Junior developers, many of whom are still forming foundational skills, aren’t using AI at the same level. Only 13% say over half of their code is AI-generated. That tells us two things: first, they’re less confident in supervising AI’s decisions, and second, they don’t yet have the judgment to separate solid output from flawed logic.
For C-suite leaders, this calls for more than just dropping AI tools into your workflow. You need experienced people guiding the integration. You also need an internal culture that treats AI as an evolving partner, not a magic wand. Investing in mentorship between senior and junior developers isn’t just a talent retention strategy, it’s how you build trust and critical oversight at scale.
In Fastly’s July 2025 survey of 791 professional developers, 32% of senior developers reported that more than half of the code they shipped was AI-generated. That’s not experimentation, that’s adoption. A senior developer in the study put it plainly: “AI will bench-test code and find errors much faster than a human, repairing them seamlessly.”
Editing and revising AI-generated code often offsets some of the anticipated time savings
Using AI in software development is no longer optional, it’s inevitable. But that doesn’t mean it’s frictionless. A sizeable portion of developers are spending too much time correcting AI-generated output. According to Fastly’s July 2025 survey, 28% of developers said they have to fix or edit AI code often enough that it erases most of the expected time savings. Only 14% say the output is ready with little adjustment. That pattern should get your attention.
The promise of AI in code generation is faster delivery and higher throughput. That’s happening, but not for everyone, and not always. When discrepancies in AI assumptions surface, whether it misinterprets program logic or generates an irrelevant implementation, developers step in, often rebuilding parts of the code. It’s not a dealbreaker for senior engineers, who know what to look for, but for many juniors, especially those still building intuition, the interruptions stack up.
The gap here isn’t just about experience. It’s about expectation. Many AI tools remain highly confident, whether or not the output is right. Developers, especially those without a deep base of experience, can end up debugging longer than they would if they had coded it from scratch. That increases the risk of performance bottlenecks and bugs slipping into production. You end up trading immediate speed for accuracy, unless you have a team equipped to do both.
For executives, the message is simple: AI tools aren’t fully autonomous. They’re force multipliers, but only when paired with robust human oversight. Shifting resources into refining workflow design, unit testing practices, and AI-specific code review frameworks is essential if you want AI to be a net accelerator. Throwing AI into teams without adjusting the quality assurance process won’t just limit benefits, it adds risk.
This friction is predictable. It’s also solvable. Organizations that succeed with generative AI will be the ones that architect around the real-world usage.
AI tools are perceived as accelerating coding speed, especially by senior developers
Experienced developers aren’t just using AI, they’re using it to go faster. According to Fastly’s July 2025 survey, 59% of senior developers said AI helped them ship code more quickly. That compares to 49% of junior developers. More importantly, senior engineers were twice as likely to report “significant” speed gains.
This isn’t about marginal improvements. These developers are accelerating output in a way that impacts product timelines. And the differentiator is clear: they can filter good AI output from bad almost instantly. That ability, to triage, adjust, and deploy faster, is something junior developers haven’t developed yet. While more than half of junior engineers did feel moderately faster, most aren’t seeing transformational impact. That’s a skill gap.
Fast execution isn’t just about writing more lines of code. It’s about increasing throughput while maintaining stability. Senior developers have already internalized what quality looks like, so when AI provides attributes, function stubs, or error handling patterns, they know if it’s viable. That reduces cognitive friction. In contrast, less seasoned developers may hesitate, verify too often, or follow flawed suggestions. That lowers velocity.
For executives and product leaders, that means you get more from your most experienced talent when AI is part of their toolkit. But to distribute that performance across the team, don’t assume the tools will do the work on their own. Training and operational alignment around AI expectations matter. If you want faster sprints, improve your teams’ decision-making around what AI should, and shouldn’t, be doing.
A junior developer included in the survey underlined this challenge: “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.” That disconnect not only slows the process, it highlights an unresolved weakness in tool-to-developer alignment. Fix that, and the gains become scalable.
AI tools enhance the overall enjoyment of coding for developers
There’s a psychological shift happening. Developers aren’t just coding faster, they’re enjoying it more. Nearly 80% of respondents in Fastly’s July 2025 developer survey agreed that AI makes the coding process more enjoyable. That’s a strong consensus. When work feels less tedious and more stimulating, engagement rises. And when engagement rises, quality and innovation often follow.
This isn’t about novelty. It’s about function. Developers are using AI to eliminate repetitive tasks, not because they want shortcuts, but because their time is better spent solving harder problems. AI isn’t replacing the work that makes engineering valuable; it’s absorbing the parts that slow it down. That shift is increasing job satisfaction, and it’s giving teams more energy to focus on complexity, design, and performance, the things that actually move the product forward.
For leadership, this matters. Enjoyment isn’t just a soft metric. It drives retention, attracts top engineering talent, and improves team cohesion. Developers who feel empowered by their tools don’t burn out as quickly. They’re less likely to churn. They also tend to have a stronger sense of ownership over the software they ship.
This stat, 4 out of 5 developers reporting a more enjoyable coding experience, should inform how you deploy AI internally. If you’re still evaluating AI based strictly on output speed or cost savings, you’re missing a key metric: how your people feel while doing the work. AI tools that improve developer experience extend the longevity and effectiveness of your teams.
Key takeaways for decision-makers
- Senior developers drive AI adoption: Seasoned developers are integrating AI into production workflows at a high rate, with 32% saying over half their shipped code is AI-generated. Leaders should empower experienced engineers to lead AI integration and mentor junior staff.
- Editing AI output eats into speed gains: 28% of developers spend significant time correcting AI code, undermining expected productivity boosts. Organizations should invest in testing standards and QA processes tailored for AI-assisted development.
- Experience dictates AI impact on velocity: Senior developers report stronger speed gains from AI use, driven by their ability to quickly assess and fix flawed output. Leaders should align AI training and tool access with developer experience levels to maximize returns.
- AI improves developer engagement: Nearly 80% of developers say AI makes coding more enjoyable by offloading repetitive tasks. Supporting AI adoption can enhance team satisfaction, reduce burnout, and support talent retention.
