When generative AI arrived, many assumed it would make the best people even better. Several rigorous studies found something different. The largest gains often went to novices.
The evidence for the novice effect
Brynjolfsson, Li and Raymond followed more than 5,000 customer support agents as an AI assistant was introduced. Productivity rose by about 15% on average.
Novice and lower-skilled agents improved by about 34%. The most experienced agents barely changed. Agents with two months' experience and AI performed like those with six months and no AI.
Noy and Zhang found the same pattern in professional writing. AI raised quality and cut time most for people who started furthest behind.
Why novices gain most
AI is trained on the work of many people, including the best performers. In effect, it passes some of their know-how to everyone.
Experts already have that know-how, so the tool adds little. Novices do not, so it adds a great deal.
The 2026 twist
Here is the paradox. Some of the jobs where AI helps novices most are the jobs that are shrinking. Stanford's AI Index 2026 found employment for developers aged 22 to 25 fell by nearly 20% since 2024.
PwC's 2026 Global AI Jobs Barometer adds a twist. In US data, AI-exposed entry roles were seven times more likely to require senior skills such as judgement and leadership.
Those roles grew 35% since 2019, while other entry roles shrank by 10%. Junior work is changing shape as well as shrinking.
Experience also shapes how well people use AI. Anthropic's March 2026 Economic Index found experienced users attempted higher-value tasks and succeeded more often.
Build this capability in your organisation
Growth Performance designs bespoke programmes that turn insight into measurable behaviour change.
The risk to expertise
If novices reach good results by leaning on AI, do they build deep judgement? Or do they produce expert-looking work without expert understanding?
That works until the AI is wrong and nobody notices. Protecting how expertise develops is part of Adaptation, where people direct, combine and check AI agents.
What this means for leaders
Speed up the early stages. Use AI to get new people to a competent standard quickly.
Keep the struggle that builds judgement. Protect some unaided practice, especially on edge cases and difficult calls.
Keep hiring and developing juniors. If entry roles vanish, your future senior talent vanishes with them.
Try this
Look at your newest people's first ninety days. Find tasks where AI gets them to a good standard fast.
Use the time saved for harder learning: judgement, edge cases and real decisions. Test one change with the Experiment Planner.