Think about two tasks side by side

What could change in this task?

AI may help prioritize amounts, dates, and missing fields in synthetic invoices. It may group similar records, but a person still needs to review the results before treating them as reliable.

Where does human responsibility remain?

A person checks whether the document belongs to a real business transaction, verifies the approval chain, and reviews exceptions. They ask the relevant person for clarification when needed.

What don't we know yet?

Document formats, approval processes, and tax rules can differ by country, organization, and period. This card is a learning exercise, not a real accounting or regulatory decision.

Transferable skills

Your first learning step

1. 10 min: Compare 8 synthetic invoices with the table records.
2. 10 min: Mark missing or conflicting fields and write why.
3. 10 min: Build a checklist and route two findings for human approval.

What could change in this task?

AI may suggest a draft summary and changes to review from synthetic spreadsheet data. It may help prepare a chart or text draft, but a person must check whether the interpretation is accurate.

Where does human responsibility remain?

A person verifies the data source and period, investigates the business context of important differences, and reviews the report before presenting it to an authorized person. They correct inaccurate or incomplete interpretations.

What don't we know yet?

Reporting templates, account definitions, and disclosure duties can vary with an organization and local regulation. Results in the learning example are synthetic and do not indicate a real financial position.

Transferable skills

Your first learning step

1. 10 min: Clean a 12-row synthetic income-and-expense table and add a source column.
2. 10 min: Prepare two simple tables and a neutral three-sentence draft summary.
3. 10 min: Compare each finding with the raw data and note two points to verify.

What could change in this task?

AI may help rank a synthetic task list and draft reminders. It may group incoming messages by topic, but a person must decide the priority and appropriate response.

Where does human responsibility remain?

A person assesses urgent or sensitive matters in context, confirms the owner and due date, and limits the sharing of confidential information. They communicate with the relevant people when tasks conflict.

What don't we know yet?

Workflows, authority boundaries, privacy rules, and record-retention periods can vary by organization and country. The scenario here is synthetic and does not replace a real office procedure.

Transferable skills

Your first learning step

1. 10 min: Enter 10 synthetic office requests with no personal information into a tracking table.
2. 10 min: Add an owner, date, status, and whether human approval is needed for each request.
3. 10 min: Identify two conflicting tasks, draft a safe reminder, and note the reason for the decision.

What could change in this task?

AI can suggest test-case drafts and edge cases. It can summarize test results and speed up repetitive checks.

Where does human responsibility remain?

A person sets the acceptance criteria, prioritizes risks, and remains responsible for the release decision. AI output is not accepted blindly; the test itself is verified to run as intended.

What don't we know yet?

The benefit depends on the tools' capabilities, the team's test automation, and security rules. Data access and approval processes can vary by country, sector, and software criticality.

Transferable skills

Your first learning step

1. 20 min: Write three acceptance tests and three edge-case tests using synthetic user records with no personal information.
2. 30 min: Run a test for a simple function; record the failed result as expected versus actual.
3. 25 min: Ask AI for two test drafts; inspect the inputs and outputs, then revise one without approving it blindly.

What could change in this task?

AI can help classify and summarize data and make possible patterns visible. It can speed up chart explanations and first-draft interpretations.

Where does human responsibility remain?

A person questions the purpose of the measurement, data quality, and context, and filters out misleading relationships. Responsibility for how findings are used and communicated remains with people.

What don't we know yet?

Results depend on the tool's data and language capabilities, the organization's data infrastructure, and access security. The same interpretation may not hold when the country, sector, or data-collection method changes.

Transferable skills

Your first learning step

1. 25 min: Flag missing and inconsistent values in a 30-row synthetic sales table containing no personal data.
2. 30 min: Create two simple summaries and one chart; write the limit that the data supports for each finding.
3. 20 min: Ask AI for an alternative interpretation; check it for sample, context, and causality, then revise it if needed.

What could change in this task?

AI can help classify support records and rank possible resolution steps. It can speed up response drafts and summaries of similar issues.

Where does human responsibility remain?

A person verifies the user's actual need and the system impact, then chooses a safe fix and the right escalation path. Protecting sensitive information and explaining the resolution clearly remain human responsibilities.

What don't we know yet?

The benefit depends on the tool's technical capability, the organization's knowledge base, and access-security arrangements. Record retention, remote-access, and approval rules can vary by country, service, and regulation.

Transferable skills

Your first learning step

1. 20 min: Read a synthetic support record with no personal information; separate symptoms, possible causes, and unknowns.
2. 30 min: In an isolated practice environment, order three resolution steps; note each step's reversibility and risk.
3. 25 min: Ask AI for a response draft; without sharing confidential information, edit it for accuracy, safety, and clarity.

What could change in this task?

AI can help organise draft ideas and suggest different ways to express them. A person checks the final text for meaning, sources and audience fit.

Where does human responsibility remain?

The editor evaluates purpose, tone, context and the effect on readers. Citing sources, checking rights and making the final publishing decision remain human responsibilities.

What don't we know yet?

Copyright conditions, the accuracy of claims and an organisation's publishing rules can vary by country, platform and content. AI suggestions may miss cultural nuances or local requirements.

Transferable skills

Your first learning step

1. 20 min: Write an 80-word draft from a fictional student-club brief; use only invented people and organisation names.
2. 25 min: Compare two factual claims in the draft with reliable public sources; clearly mark any wording you cannot verify.
3. 30 min: Review the text for headings, short paragraphs, link text and screen-reader-friendly structure; do not use real customer data.

What could change in this task?

AI can help suggest variations on an idea and rough layouts. A person reviews the output for originality, rights, brand context and accessibility.

Where does human responsibility remain?

The designer evaluates visual hierarchy, cultural meanings and the real setting in which the work will be used. Preparing a usable final file and checking rights remain human responsibilities.

What don't we know yet?

Licences for visual resources, brand rules and the cultural meaning of symbols can vary by country and organisation. AI output may create similarity, misrepresentation or accessibility problems.

Transferable skills

Your first learning step

1. 25 min: Sketch two black-and-white layouts from a fictional school-event brief and compare the headline with the call to action.
2. 30 min: Verify the one factual claim on the fictional poster with a reliable source; label it “unverified” if you cannot confirm it.
3. 20 min: Check the draft for colour contrast, text size, alternative text and non-colour cues; use only your own or royalty-free elements.

What could change in this task?

AI can help sort reply drafts and try different tone options. A person checks personal data, organisation policy, commitments and sensitive situations.

Where does human responsibility remain?

The communications specialist decides what can be said while considering the other person's needs and feelings. Protecting privacy, checking authority and escalating to a person when needed remain human responsibilities.

What don't we know yet?

Organisation policies, consumer rules and sensitive wording can vary by country, sector and culture. AI may produce a wrong, overconfident or context-inappropriate reply.

Transferable skills

Your first learning step

1. 20 min: Write three questions for a fictional student club and create two reply tones that contain no personal data.
2. 25 min: Verify two factual claims in the replies with reliable public sources; if you are unsure, state the uncertainty clearly.
3. 30 min: Rewrite one reply with short paragraphs, clear headings and an understandable next step for accessibility; add a handoff to a person for a sensitive situation.