Slug: task-portfolio-ai-era-career-evidence
Tags: Future of Work, building career, AI
Meta description: Build a career task portfolio with clear inputs, visible AI assistance, human corrections and evidence of the work you can deliver.
“I know how to use AI” is becoming an incomplete description of a skill. A prospective employer or client still needs to know what you can produce, how you check it and which part of the result depended on your judgement.
A task portfolio answers those questions through a small set of work examples. Instead of presenting a list of tools, show a problem, the source material, the assistance you used, the decisions you made and the finished result.
Begin with work, not a prediction about your job title
The ILO–NASK index published in May 2025 assesses potential generative-AI exposure at the task level. It argues that transformation is the more likely overall effect because many occupations contain work requiring human input. Exposure is not a count of future redundancies, and the study does not guarantee an outcome for an individual career.
A practical response is to inspect the tasks in the work you want to do. Which ones require accurate synthesis, communication, prioritisation, verification or collaboration? Which can you already perform, and where could AI assist a part of the process?
Choose three representative tasks
Select a small set that reflects the role. A communications candidate might prepare a sourced explainer, adapt it for two audiences and build a simple content plan. An operations candidate might compare proposals, identify missing information and prepare an escalation record.
Use public, invented or explicitly authorised materials. A portfolio should not expose private employer documents or customer information. If you use a fictional scenario, label it plainly so the reader does not mistake it for a paid client project.
Give each example a six-part record
| Part | What to show |
|---|---|
| Problem | The task and the intended user or decision |
| Inputs | The permitted sources and relevant constraints |
| AI contribution | What the tool drafted, extracted or suggested |
| Human judgement | What you checked, changed, rejected or clarified |
| Result | The final deliverable and acceptance criteria |
| Limitations | What the example does not establish |
The record can be short. A clear page with links to the work is often more useful than a long description of every prompt. Make the reader’s task easy: they should be able to see the deliverable and understand your decisions.
Show one correction that mattered
A polished final file gives little information about how you handle mistakes. Include a before-and-after passage or a brief error record. Explain the source of the issue, the check that caught it and the change you made.
For a research explainer, you might show that a generated draft described an association as a cause. Point to the original study wording and explain the correction. For a proposal comparison, you might show that an important service was “not stated”, then record the question you would ask the supplier.
These examples are hypothetical. In your own portfolio, choose a correction you actually made. Do not manufacture a dramatic AI failure to make your role seem more valuable.
Measure carefully, or leave the number out
If you claim a time saving, preserve the method. Compare the same task under the same quality standard and explain how you measured the time. A rough personal estimate should be labelled as such.
Do not imply that a sample produced revenue, improved a team’s performance or reached a large audience unless you have evidence. A simulated project can still demonstrate useful capability without an invented commercial result.
Make tool use transparent
State which assistance you used and for what purpose. The relevant disclosure is functional: “AI suggested three outlines; I selected and revised one, checked the claims and wrote the final recommendations.” That says more than placing a provider’s logo on the page.
Also show what you can do without the tool. Be ready to explain the source, the reasoning and the acceptance criteria in a conversation. If you cannot explain a result you submitted, the portfolio has revealed a learning need.
Use the portfolio to direct your next learning step
Review the three examples for repeated gaps. Perhaps you can draft well but need stronger source checking. Perhaps you can analyse a document but struggle to explain the decision to another person. Let the work identify the next skill to practise.
Keep an earlier version and revisit one task after practice. The difference can show improvement more clearly than adding another certificate or another tool to a list.
A practical first week
Choose one task and write its acceptance criteria. On the next session, produce a first attempt using approved sources. Then use AI for one bounded form of assistance. Spend another session checking and revising the result. Finish by writing the six-part record and asking someone to explain what they think it demonstrates.
Their feedback can reveal a presentation gap. If they cannot see your contribution, improve the record. If they misunderstand the result, revise the explanation. Treat the portfolio itself as a piece of communication work.
Business + Psychology + AI
Business: clients and employers can assess relevant work. Psychology: concrete examples can make progress visible without pretending uncertainty has disappeared. AI: assistance becomes part of a transparent process that you can describe and evaluate.
Start with one honest example. Use our AI run-sheet guide to record the workflow, and our work-transfer guide to connect learning with a real task.
Reference: ILO–NASK: Generative AI and Jobs, a refined global index of occupational exposure (20 May 2025).
Featured image: conceptual illustration created with AI for MaryChuks.com.
Discover more from Marychuks.com AI, Psychology, Business & CreativeVerse
Subscribe to get the latest posts sent to your email.
