Recognition can support useful AI adoption, but it cannot guarantee cooperation or sustained use. Recognize a contribution that made the work better: a careful check, a reusable improvement or help that enabled someone else to do the task. Make the contribution and its beneficiary visible, share credit, and check whether the programme is encouraging helpful work or competition for attention.
For a company with several functions or branches, the question is also who has a fair chance to qualify. Someone with frequent drafting tasks, an approved AI tool and time to experiment has different opportunities from someone whose work rarely permits AI. A company-wide usage ranking hides those differences.
Recognize the contribution you want repeated
A thank-you tells people what the organization notices. “Most AI messages this month” rewards activity. “Caught an incorrect collection promise and showed colleagues how to check it” describes useful work that others can understand and repeat.
There is evidence that recognition can encourage effort. In a field experiment reported by Christiane Bradler and colleagues, mainly student workers doing a three-hour data-entry job received an unexpected public thank-you after two hours. Subsequent performance improved under several recognition conditions, including selective recognition. This was a short, one-time task, not an AI rollout or continuing collaborative team. It challenges the claim that public or selective praise necessarily harms performance, without establishing which approach your programme should use.
The distinction matters when the reward measures something easy to increase. In a June 2026 X post, AI adviser and course provider Allie K. Miller reported hearing privately from employees who fed full novels to corporate AI systems to raise their token-usage numbers. These are anonymous accounts relayed by Miller, not independently verified company cases. They describe a plausible failure of the criterion: the count can rise without useful work changing.
In a reply, Yann Kronberg argued for finding the changed workflow first and checking whether usage followed it. That is a useful starting question for recognition: what did someone improve, and how can the person who benefited explain the difference?
Treat public rankings as a separate design decision
Recognizing someone and ranking everyone are different choices. A shared story can explain a useful practice without placing employees above and below one another.
Even rankings are not interchangeable. In a March 2026 account from the researchers' university, a laboratory study involving 135 participants compared leaderboards for recognition given, recognition received, both, or neither. Participants could help others at a cost to their own payment. Ranking recognition given increased helping compared with no leaderboard; ranking recognition received, or both measures, reduced it.
The study concerns laboratory helping, not sustained AI adoption at work. Its useful warning is that what a leaderboard ranks can change the response. The result is not a reason to launch a praise-giving contest without checking whether the thanks remain specific, sincere and useful.
Ownership can complicate the picture too. In a September 2025 Reddit discussion, a self-described team lead reported a task-completion leaderboard introduced by the CEO. An intern then reassigned a documentation task from a full-time colleague and delivered it at 3am. The colleague felt the work had been taken from them. The author's follow-up shows disagreement about whether the handoff had been agreed.
This is one person's account, not an AI deployment or proof that the leaderboard caused the conflict. Another Reddit user emphasized assignment authority: management needed to clarify who could move the work. A recognition programme needs that rule too. Completing a visible task should not erase another person's contribution or bypass its owner.
Give the checker and the helper credit
Suppose Jamie Lee handles service reception for a bicycle retail chain. Jamie uses an approved AI assistant to draft a collection notice from permitted, non-sensitive repair facts. Mechanic Sam Taylor checks it against the repair ticket and spots that a replacement part is still pending. The bicycle is not ready for collection.
Jamie corrects the draft before it reaches the customer. Sam explains the check so Jamie can apply it to the next notice. Recognizing only the person who produced the draft would leave out the judgment that prevented an incorrect promise. Recognizing only Sam would miss Jamie's correction and willingness to learn.
Use the actual work to decide what deserves acknowledgment. These are possible criteria to adapt, rather than a validated award scheme.
| Contribution | Evidence worth discussing | Credit safeguard |
|---|---|---|
A useful AI-assisted draft | The task owner can explain how it helped and what needed review | Name the drafter and significant reviewers |
A material error caught | The source, discrepancy and correction are clear | Recognize checking even when it reduces apparent speed |
A practice taught | A colleague can apply the check to another relevant task | Credit both the teacher and the colleague's own work |
An improvement shared across branches | Another branch can explain what it reused and adapted | Keep the originating and adapting contributions visible |
A message to Jamie and Sam might say: “Thank you for checking the collection notice against the repair ticket and catching the pending part before we contacted the customer. Showing that check at reception will help us handle the next notice accurately.”
That message explains the contribution without awarding points for how much AI either person used. It also leaves the service owner responsible for the customer communication. Recognition does not change approval or assignment rules.
Check who can participate and how they want to be thanked
Before opening nominations, ask whether employees have the conditions to qualify:
- Permitted work: Does the role have relevant tasks where AI use is allowed?
- Access and time: Can the employee use the approved tools and practise during work?
- Visibility: Can quieter checking, shift work and help behind the scenes reach the people choosing recognition?
- Credit: Can reviewers and collaborators be named without one person claiming the whole result?
If some employees cannot participate, change the conditions or compare contributions within appropriate work contexts. More praise will not resolve an unavailable tool or an unsuitable task. Use the guide to distinguishing inability, reluctance and lack of opportunity before interpreting low participation as poor motivation.
Ask how the person wants to be thanked. A private, specific conversation may be welcome when a public spotlight is not. With agreement, a team story can share the practice and credit its contributors. Keep sensitive customer, employee and business details out of nominations and shared examples.
Recognition also needs a boundary. Occasional thanks for helping should not quietly become an expectation that someone trains every branch alongside their existing job. When responsibilities expand, review the time, support and role. The performance-system guide connects useful adoption to the expectations and conditions of work.
Review helpful work rather than the volume of praise
Try a small recognition pilot in a defined work area before extending it across the company. Explain the criteria and invite examples from contributors and beneficiaries. Treat nominations as starting points for a conversation, not a new automated performance score.
At the review, ask:
- Are people sharing a useful check or method that colleagues can actually apply?
- Are important errors and sensible decisions not to use AI still being reported?
- Are contributors helping across teams, or protecting visible tasks to secure credit?
- Are the same roles receiving attention because their work is easier to observe?
Inspect a few work examples and speak with people who received, gave and did not receive recognition. Record changed access, workload or training alongside the observations. A rise in nominations shows more nominations; it does not establish that recognition caused better adoption. Quality, useful help and employee experience need their own evidence.
If the pilot encourages task grabbing, vague praise exchanges or hidden mistakes, revise the criterion and credit rules before expanding it. The people and change guide places recognition alongside learning, management and trust. Recognition is one part of those conditions, not a substitute for them.
Questions and answers
Should you use an AI adoption leaderboard?
Decide what the leaderboard would rank and why comparison is necessary. A ranking of token usage or prompts cannot establish useful work; a ranking of praise received can reflect visibility as well as contribution. Start with specific work stories and shared credit, then evaluate any proposed ranking against the behavior you want.
The comparison of public ranking choices explains why even different recognition leaderboards should not be treated as one intervention. The evidence here does not establish a universally effective AI adoption leaderboard.
What should an AI recognition message say?
Name the task, the specific contribution and who benefited. Include significant collaborators, and describe what others can learn from the work. For example, thank a bicycle-service receptionist and mechanic for checking a collection notice against the pending repair status before contacting the customer.
Use the contribution and credit comparison to choose evidence for the message. Confirm that the employee is comfortable with the audience and that the example does not disclose sensitive information.
Can you recognize an experiment that did not work?
Yes, when the contribution was a careful test, a useful check or a clear, relevant lesson that others can use. Explain the conditions and what changed, rather than praising failure or implying that an unsuitable AI output was successful. Recognition can also acknowledge a justified decision to stop using AI for a task.
Ask the work owner to check the lesson and its usefulness to others. The criteria for useful contributions help keep acknowledgment tied to work rather than novelty or activity alone.
How do you know whether recognition is helping AI adoption?
Look for useful work, shared help and fair participation, alongside employees' experience of the programme. Review task examples and speak with contributors, beneficiaries and people who receive little attention. Counts of nominations, tokens or awards are activity measures; they do not establish quality or causation.
Use the pilot review questions, record other changes in access or support, and revise the programme if it encourages credit disputes or hidden errors. A small local pilot can inform a decision without proving a lasting company-wide effect.
