When an AI-use contest comes to work, what does it teach people to do?
SHRM’s 2026 survey of more than 5,000 workers says AI competitions are among the approaches workers rate most effective at driving adoption. That is not quite the same as helping people do better work. A contest gives the company an easy scoreboard: prompts, projects, usage. It gives the person with a customer deadline a reason to try the tool where it is least ready, or to keep using it after the answer needs more checking than the task did. If a team runs one, it should reward a useful no as readily as a flashy use: the job the tool was not used for, the mistake caught before it spread, the manual step that disappeared for good. Otherwise it is a scavenger hunt with procurement behind it. What would keep an AI-use contest from becoming a contest in looking busy?
Comments
Give each team one recurring task and keep three columns: time to a finished result, reviewer changes, and work that comes back. Count a “no use” as a result too. A contest earns its place when a boring job actually disappears, not when people learn to manufacture demos for the scoreboard.
A ‘no use’ column only works if the manager has to read it before asking why the adoption number slipped. Otherwise it becomes a polite trash can for objections while the contest keeps rewarding demos. Give the team a short list of tasks the tool is not for, and let that list count as a result.
There is another failure mode: people learn to use the tool where it is easiest, then the manager reads a high participation rate as readiness for customer work. Make every winning entry name one task it should not be used for yet and the person who takes the first exception. That makes limits part of the result, not an objection someone has to raise after the contest.
The survey supports a narrower claim: workers who reported adoption strategies rated competitions as effective for engagement. It does not show that a contest improved customer work, reduced checking, or held up after the prizes ended. Ivy’s rule would make the next survey more useful: ask people to name the task they would still keep away from the tool, then check six weeks later whether that boundary held. A contest that raises use while quietly eroding judgment is teaching the wrong habit.