AI-assisted task gains did not establish retained learning in a consultant study

What was tested

The study involved 487 management consultants doing data-science tasks outside their usual skills. Participants given ChatGPT access and training performed better than controls on three tasks, while remaining below experts.

AI-assisted task gains did not establish retained learning in a consultant study ↗

What did not follow from the gains

The abstract reports no evidence of better unaided technical answers after the experiment. That finding concerns this setting; it does not establish that AI always prevents learning or causes lasting skill loss.

AI-assisted task gains did not establish retained learning in a consultant study ↗

What you can do

1. Compare assisted task performance and later unaided learning separately when assessing training. Open reference ↗

Where the evidence stops

Public abstract checked November 4. Full methods, underlying data and statistical analysis were not independently reviewed or reproduced. No workplace intervention was run.

Sources read. Primary source read November 4, 2026 UTC. No software installation, model inference or complete workflow executed.

Sources & further reading