Proof of Work
How to show your contribution to AI-assisted work
When AI helps produce your work, the finished output may leave your contribution unclear. Use four questions and a worked example to explain your decisions and evidence.
Quick answer
Show the task, what AI helped with and one decision you made. Connect that decision to the evidence you checked and the change you made to the work. This helps someone understand your contribution. It also helps you see which parts of the task you could hand to AI and which need your attention.
The finished work can leave your contribution unclear
You've used AI to help create a comparison, a brief or an analysis. The result looks ready to share. But someone reading it may not know which parts AI produced, what you checked or why you chose the final recommendation.
That makes your contribution harder to explain. It can also leave you unsure about what to hand over to AI next time.
A short account of one important decision can make both clearer. Start with the work, then show what you did to make it useful.
Show the decision that changed the result
Consider Maya, an operations professional comparing software plans for a 12-person customer-support team. She is working with finance and the team lead to recommend which plan to try.
This is a fictional example. Maya, the prices and the work described below are invented for illustration.
Both plans appear to cover the team's basic needs. Maya asks AI to organise the supplied plan information into a table and draft a recommendation. The first draft recommends Plan A: USD 18 per person multiplied by 12 people gives USD 216 a month.
Maya checks the plan terms. Plan A requires a minimum of 20 seats, even if only 12 people use it. Here is the corrected comparison:
| Plan | Monthly price per seat | Minimum seats | Minimum monthly cost |
|---|---|---|---|
| A | USD 18 | 20 | USD 360 |
| B | USD 25 | 12 | USD 300 |
She corrects Plan A's cost and changes her recommendation to a trial of Plan B. The team still needs to check whether it fits their workflow.
The useful output is a corrected comparison and a trial recommendation. Maya's contribution becomes visible when she shows the minimum-seat condition, the corrected calculation and the decision it changed.
Explain what you did
A useful output meets a work need. Evidence of a skill helps someone understand what you did to produce it.
Maya's comparison helps the team choose a trial. Her documented correction shows how she checked a commercial constraint, validated a calculation and revised a recommendation. AI helped organise the information; she checked the terms and made the decision.
To explain your own contribution, answer four questions:
- What task were you trying to complete?
- What did AI help with?
- What important decision or change did you make?
- What evidence supports that decision, and where can someone see it?
Combine the answers into a short paragraph. For Maya's fictional example, it could read:
I compared two software plans for a 12-person support team. AI organised the supplied information and drafted the first recommendation. I checked the minimum-seat terms and corrected Plan A's monthly cost from USD 216 to USD 360. Plan B was USD 300 for 12 seats. I recommended a trial of Plan B, with workflow fit still to check. The corrected table and plan clauses show the basis for my decision.
The note points to the relevant evidence: the terms, the original calculation and the corrected comparison. A reader can follow the decision without needing the entire chat history.
Keep the result accurate. Maya recommended a trial; the example does not show that the team bought Plan B or saved money. In your own work, include measured results when you have them and describe proposed changes as proposals.
Use the same approach in other work
Your contribution does not have to be a correction. It might be choosing an approach, comparing options or deciding which suggestion to develop.
Here are three hypothetical examples:
| Work | Contribution to explain | Evidence to show |
|---|---|---|
| Marketing brief | You changed the proposed audience after checking customer research | The relevant research excerpt and revised brief section |
| Software change | You rejected a suggested fix after reproducing the failure | The reproduction, test output and revised change |
| Finance analysis | You corrected an assumption before updating a forecast scenario | The assumption's source and revised calculation |
Explain why the evidence mattered. Naming the change tells someone what happened. Connecting it to a finding helps them understand your judgment.
Decide what to hand to AI next
Maya's task contains several kinds of work: organising plan details, checking terms, comparing costs and choosing a trial. Looking at these separately helps her decide where AI could be useful and where she needs to stay involved.
You can do the same with a task you perform regularly. Where could AI reduce repetitive work? Where could it help you compare more options or improve an analysis? What decisions remain yours to make?
This is where guided practice helps. Altiv.AI Plays provide structure, guidance and content updated as tools and working practices evolve. Each Play starts with a practical task and helps you explore how AI could make the work faster, improve the result or make a new approach possible. You examine evidence, make decisions and create something useful along the way.
If you're deciding where to begin, the Altiv.AI AI Readiness Score helps you examine tasks in your role through three diagnostic categories: automated, augmented and human-only. Use those categories to consider what AI could handle, where it could support you and what should remain with people in your working context. They are a starting point for choosing what to practise, not permanent predictions about a task or proof of your skill.
Choose what to practise next
Explore tasks in your role with Altiv.AI's AI Readiness Score.
Get your AI Readiness Score →
