AI-Assisted Promotion Case: Prove, Disclose, Defend

You are about to submit a promotion case, and one key deliverable was built with AI. The panel will not ask whether you used a tool; it will ask whether you can own the result. Use.AI's survey of 9,684 working adults in Latin America, the EU, Canada, the UK, and the US found that 52% felt AI made them seem more experienced than they were. In that survey, 39% had turned in AI-assisted work without flagging the help.

The survey also shows why panels are wary: 64% used AI to finish work they said they could not complete alone. Nineteen percent credited AI-assisted work with helping them get promoted. The Work AI Index 2026 counted 69% of AI users as admitting to botshitting at work: shipping AI-generated work that has not been reviewed, is not fully understood, or cannot be defended if challenged. The same index found that 41% of workers are shipping AI outputs they cannot explain. A panel can ask the same question in a meeting: walk me through the part you did not write?

Promotion case evidence should show judgment, not novelty. Panels want to know whether the person can operate at the next level in the org chart without a safety net. A deliverable built with AI must show that your contribution was the decision-making around it, not the prompt.

Disclose the help that changes the evidence

Minor AI help that does not shape the main deliverable may not need a formal note. Core AI help, where the tool shaped the outcome, should be known before the panel weighs the result. Ihor Herasymov, co-founder and CEO of Use.AI, argued that useful policy should center on material AI help, verification, and accountability, while leaving the submitter responsible for understanding and defending the work. In a promotion case, material means the assistance shaped the outcome, the method, or the quality the panel is judging. When AI produced the main analysis or recommendation, say so.

The case must show judgment, verification, and defense

The checklist below is the part you can test in a minute. Each item should be true before you submit the case, not after a panel asks for it.

• Disclose material AI assistance. State where AI helped, what it did, and why that does not change your ownership of the result.
• Document independent judgment. Show the choices you made: what you rejected, what you changed, and what you would do differently if the assumptions shifted.
• Show verification. Include the checks you ran, the data you compared, and the places where you corrected the output.
• Prepare to defend the output. Be ready to explain the method, the limits, and the next step without relying on the tool to answer for you.

Specificity is what makes disclosure useful. A statement that AI helped draft an early version of a work product is more useful than a bare statement that AI helped. The evidence should show whether the assistance touched the part of the job that proves readiness for the next level.

Independent judgment is the part that separates a contributor from a supervisor. The case should show how you ranked AI-generated options, what you checked before trusting an AI summary, and the business logic you added, tested, or removed in an AI draft. The choices are the point.

Verification is where the case becomes defensible. A panel may not need a full audit, but the case must show evidence that you did not accept the first answer. That can be a comparison against source data, a sanity check on edge cases, a review of assumptions, or a note about what the output could not see.

Make the panel's question easy to answer

Before the meeting, read your case as a skeptic. Vague labels invite follow-ups: AI-generated insights require what you did with them, improved process demands the change and the measure, and led initiative points to the moment where your judgment appeared when the work was uncertain.

Do not hide the parts that are weak. A panel will notice a gap faster than you expect, and a disclosed limitation can become proof of judgment. Show what you added to useful but incomplete output, how you caught a wrong spot, and why you trusted strong output.

Bring the checklist, the disclosure, and the verification notes into the meeting. Let the evidence answer the hard question before it is asked.