A McKinsey study tried to measure how much faster AI makes developers. The result surprised everyone: junior developers with less than a year of experience actually took 7 to 10 percent longer to complete tasks with AI tools than without them. Meanwhile, experienced developers saw productivity jump by 35 to 45 percent. Same tools. Same AI. Opposite results.
That finding lands on something we’ve seen firsthand across hundreds of projects: AI doesn’t replace experience. It amplifies it. And if you’re planning a web project in 2024, that distinction should shape every decision you make about the team you hire.
TL;DR
McKinsey research found that junior developers actually took longer to complete tasks with AI tools, while experienced developers saw 35 to 45 percent productivity gains with the same tools. The difference comes down to judgment — experienced developers evaluate AI output, catch subtle bugs, and know when to reject suggestions that look right but aren’t. AI coding tools amplify existing skill. They don’t create it. A strong team with AI is powerful. A weak team with AI ships bugs faster. When choosing a development partner, ask about their team’s experience and quality processes, not just which AI tools they use.
The data tells an unexpected story
When McKinsey’s technology practice tested generative AI with over 40 developers across the United States and Asia, the headline numbers were impressive. Developers could complete coding tasks up to twice as fast. Code documentation took half the time. New features got built in nearly half the usual timeframe.
But those gains weren’t distributed evenly. The developers who benefited most already knew what good code looked like. They used AI to move faster through work they understood deeply. Experienced developers saw code development speed jump 35 to 45 percent and refactoring speed increase 20 to 30 percent.
Junior developers? A different picture. With less than a year of experience, some actually performed worse with AI assistance. Tasks took 7 to 10 percent longer. The reason is revealing: AI generates code that needs to be critiqued, validated, and improved. Without the skills to evaluate what AI produces, you’re not getting a productivity tool — you’re getting a source of confusion.
This mirrors what we discussed in our earlier post about how AI speeds up development without cutting corners. The speed is real, but it depends entirely on who’s holding the controls.
Why experienced developers get more from AI
Think about giving a power drill to a master carpenter versus someone who’s never built anything. The tool is identical. The results will be completely different. We explored this analogy in our first article in this series, and the McKinsey data confirms it with hard numbers.
Experienced developers bring three things to AI-assisted work that juniors don’t have yet:
Pattern recognition. A senior developer has seen thousands of code patterns across dozens of projects. When AI suggests a solution, they instantly recognize whether it follows best practices or introduces a subtle anti-pattern. This comes from years of building, breaking, and fixing software.
Architectural judgment. AI generates code that works in isolation but is weaker at understanding how it fits into a larger system. An experienced developer knows that a function which passes every test can still be wrong if it doesn’t handle edge cases or scale as the business grows.
The instinct to say no. GitHub’s research shows that developers using Copilot complete tasks 55 percent faster. But speed only matters if you’re going in the right direction. Experienced developers reject AI suggestions regularly because they recognize when something is subtly wrong. Junior developers tend to accept more suggestions because they can’t tell the difference.
The junior developer paradox
The developers who need the most help are the ones least equipped to use AI effectively.
When a junior developer asks AI to write a database query, the tool produces something that works. It returns the right data. But it might not handle SQL injection properly, account for performance at scale, or anticipate edge cases that an experienced developer catches instinctively.
The Stack Overflow 2023 Developer Survey found that 44 percent of developers were already using AI tools. But developers were split on whether they trusted AI output. Experienced developers were the most cautious — they knew from experience what could go wrong. That caution protects your project.
There’s also a learning concern. When juniors lean heavily on AI, they may skip the struggle that builds deep understanding. It’s like using a calculator before understanding multiplication — you can get answers, but you can’t spot when the answer doesn’t make sense.
What this means for your next project
If you’re choosing a development team, these findings should change what you look for:
Ask about team experience, not just tools. Every agency uses AI now. The differentiator is who’s using those tools. A team of experienced developers with AI delivers dramatically better results than juniors with the same tools. We covered the specific questions worth asking in our post about 5 questions to ask your web agency about AI.
Look for strong quality assurance processes. AI-generated code needs more review, not less. The best teams pair AI-assisted development with rigorous testing and code review. The McKinsey study specifically noted that quality assurance becomes more important when AI is part of the workflow.
Value judgment over speed. The fastest development isn’t always the best. As we explored in our post on AI and development speed, the goal isn’t just to build faster — it’s to build better, faster. That distinction matters when code needs to run reliably for years.
Invest in teams that invest in their people. Agencies that mentor junior developers, maintain code review standards, and pair less experienced developers with senior engineers produce better work. Strong development teams don’t just use AI — they use it wisely, with guardrails built by experience.
Frequently asked questions
Can junior developers use AI coding tools effectively?
They can, but with caveats. McKinsey’s research found that developers with less than a year of experience sometimes took longer with AI tools. Junior developers benefit most when paired with experienced mentors who help them evaluate AI output and learn from the process rather than just accepting suggestions.
Does AI make experienced developers less necessary?
The opposite. AI makes experienced developers more productive and more valuable. The McKinsey study showed 35 to 45 percent productivity gains for experienced developers — gains that depend on judgment only experience provides. AI handles routine code generation, freeing experienced developers to focus on architecture, quality, and complex decisions.
Should I ask my development agency about their developers’ experience levels?
Absolutely. Developer experience determines whether AI tools help or hurt project quality. Ask how many years of experience their team averages, how they pair junior and senior developers, and what quality review processes they use on AI-generated code.
What’s the ideal balance of AI tools and developer experience?
The best results come from experienced developers using AI as one tool in a workflow that includes human code review, automated testing, and architectural oversight. AI handles speed; human experience handles judgment and quality.
Will AI eventually make developer experience irrelevant?
It’s unlikely in the foreseeable future. Software development requires understanding business context, user behavior, system architecture, and edge cases that AI can’t reason about independently. Developers who understand these dimensions will remain essential — and AI will continue to make them more effective.
The bottom line: invest in people, not just tools
AI is transforming how software gets built. But the transformation isn’t about the tools — it’s about what skilled people do with them.
The McKinsey data makes this concrete: the same AI tools that make experienced developers 35 to 45 percent more productive can actually slow down less experienced ones. Your development team’s expertise isn’t a nice-to-have. It’s what determines whether AI helps your project succeed or introduces risks you can’t see.
When evaluating a development partner, look past the AI marketing. Ask about the team. Ask about their experience. Ask about their quality processes. The agencies that invest in their people — not just their tools — deliver the best results.
Ready to work with a team that brings both deep experience and smart AI integration? Let’s talk about your project.
