Make your AI promotion case about judgment, not output

Your review form now says AI fluency. Your manager says the bar is higher, and the org chart says the next level owns decisions, not tool clicks. If your promotion review measures AI use, the strongest case shows how you used AI to make better decisions, not how much you used it.

AI promotion criteria can sound like a usage test. At Accenture, CEO Julie Sweet said in March that promotion depends on doing the work the company says it now carries out with AI. Disney, Meta, JP Morgan and KPMG have introduced AI leaderboards that rank employees' use of language-model tools. Coinbase fired engineers who did not complete AI training that CEO Brian Armstrong had requested. The pressure is real, but a promotion panel should reward judgment, not volume.

You are trying to prove you can be trusted with more. A promotion case that reads like a tool log makes the reader work for the judgment; a case that reads like a decision makes the reader’s job easy.

The review language is the real promotion bar

Start with the exact words in your review. “AI use” is weaker than “AI fluency.” “Adoption” is weaker than “judgment.” When the criteria mention training, show what changed after the training, not that you completed it; when they mention output, show where output was checked, corrected, or redirected.

Gi Group's July poll of 1,881 UK jobseekers found that 75% would not be deterred from applying to employers that added AI proficiency to individual performance reviews. The poll suggests the market is normalizing the practice, so your case needs to be clearer than the average applicant’s.

Judgment evidence beats usage totals

Henley Business School's Kamila Miller said making AI usage a KPI measures obedience rather than judgment. A promotion panel does not need to know you clicked the model often. It needs to know you can separate a useful answer from a confident mistake.

Run this checklist before the next review meeting. Each item pushes the case toward the thing the next level actually owns: decisions.

• Locate the AI-fluency language: point to the exact phrase in your review criteria that mentions AI use, fluency, adoption, or judgment.
• Collect three judgment examples: name one moment where you framed the problem, one where you caught an AI error, and one where you explained a trade-off.
• Write the case as judgment evidence: check whether your draft leads with the decision, the risk, and the outcome before the tool.
• Ask your manager to calibrate it: get a calibration answer on whether the case meets the promotion bar or only shows activity.

Vague criteria invite vague cases. The review says “AI fluency,” so do not write “I used AI.” Write what fluency changed: a faster first draft, a cleaner comparison, a sharper question to the data.

Collecting examples is the core. A good example has a before and after. Before, you had a problem. After, you had a decision. The AI step is in the middle, not the title.

Writing the case as judgment evidence keeps it from becoming a tool review. The panel is evaluating you, not the model. A draft that praises the tool should be rewritten as a decision record.

Asking for calibration is the step most people skip. Your manager can test the draft against the bar and the org chart. Request calibration, not praise. A calibration answer that calls the case activity still leaves time to fix it.

Calibrate before the panel sees the case

A promotion case predicts that you can handle the next level without the current level’s support. The AI language in the review asks for proof of control, not just participation.

When your company tracks usage, do not fight the metric. Translate it. A high usage total can support a judgment example, but it cannot replace one. The panel should leave with a clear answer: this person can use AI well enough to make the next level’s decisions.

The legal context is also changing: UK employment lawyers said measuring staff by AI use is lawful. UK unfair-dismissal rules change in January 2027, letting employees bring claims after six months of service and within six months instead of three. Ask the manager to mark the example that meets the bar and the one that still reads as activity.