Source: Wang et al. (2024) — “Friend or Foe? Teaming Between Artificial Intelligence and Workers with Variation in Experience,” Management Science
Experience is one of those words that sounds simple until you actually measure it.
Someone can have experience doing a specific task many times. Someone else can have seniority, broad organisational knowledge, and the confidence of a person who has survived many meetings.
Those are not the same thing.
This paper studies how workers with different types of experience team with AI, using a field study in medical chart coding. The finding is useful because it separates narrow task experience from broader seniority.
AI helped workers with greater task-based experience. That makes sense. If you know the work deeply, AI can become a fast assistant. You can spot when it is helpful, catch mistakes, and use it to move faster.
But senior workers did not necessarily gain as much. The lower productivity lift was linked to lower trust in AI, not seniority itself.
I find this very believable.
People with broad experience may have stronger habits, stronger judgment, and more reasons to question a system. Sometimes that scepticism protects quality. Sometimes it blocks useful support. The line is not always obvious.
This complicates the lazy idea that AI helps juniors and threatens seniors.
Reality is messier.
AI may help people most when they have enough task knowledge to use it well and enough openness to actually use it. Too little knowledge, and they may not know what to check. Too little trust, and the tool sits there like an expensive ornament.
My takeaway: AI training should not treat experience as one thing. Task experts, senior employees, and newcomers may need different support. The goal is not blind trust. It is calibrated trust, which is much less catchy but much more useful.