Does AI make everyone equally more productive, or mainly help beginners while experts gain less — or even slip?

This explores whether AI's productivity boost depends on who is using it and what they're trying to do. The corpus doesn't give one simple answer. Who gains depends on how much a worker already knows, and on whether they're doing familiar work or learning something new. The clearest data point comes from customer support. Across 5,172 agents, AI assistance raised issues resolved per hour by 15% on average. Most of that gain went to less experienced agents, who got faster and better. The most experienced agents barely sped up and saw slight drops in quality Does AI assistance help less experienced workers most?. In that setting, AI works like a way of passing expert knowledge down to newcomers. It helps less the closer you already are to the top.

High expertise can even turn the gain negative. In a randomized trial, experienced open-source developers working in codebases they knew well took 19% longer with early-2025 AI tools. Before starting, they had predicted a 24% speedup Do AI coding tools actually speed up experienced developers?. Part of the reason is that deep familiarity leaves little room for AI to add anything, while checking unreliable suggestions still costs time. That fits a broader pattern. A survey of 3,200 AI users found that almost 40% of time saved is lost to fixing and verifying outputs Where does AI's time savings actually go in practice?. A 535-person study found that people working with an LLM captured only about half of the model's accuracy gains, ending up worse than the better of the two working alone Why does assisted accuracy capture only half the LLM gain?. Turning AI output into real productivity is a skill of its own, and it doesn't happen automatically.

The twist is that 'less experienced' doesn't mean 'learning.' The support agents' gains came from doing a job they had already been trained for. When workers used AI to pick up genuinely new skills, the productivity gains disappeared and learning suffered When does AI actually boost worker productivity?. One likely reason: polished AI output feels easy to read, and people take that ease as a sign of their own competence, even though they didn't produce the work Does processing ease mislead users about their own competence?. So a novice can feel more capable while building less real skill. This may help explain why perceived gains run ahead of measured ones, both for individual developers and for the executives surveyed in Do AI productivity gains feel larger than they actually measure?.

Over time, skill level isn't fixed. AI changes it. Interviews with knowledge workers found four different outcomes: some skills grew, some held steady, some faded, and some changed in value. Which one happened depended on how each person used the tool How does generative AI actually change worker skills?. Narayanan and Kapoor point to why. AI mostly shrinks the 'execute' part of knowledge work, while deciding what to do and delivering the result stay the same or grow Does AI really compress all layers of knowledge work equally?. That suggests a takeaway you might not expect. AI helps novices most at execution, the part it is absorbing. The judgment that experts already have, and that AI-assisted learners may never build, is the part that keeps its value.