By Ronnie Mbuqe
Drafting is only one part of delivering a useful report. Someone must choose the question, supply accurate information, interpret consequences and take responsibility.
A junior analyst can spend less time formatting and more time checking assumptions, but only if the organisation redesigns the work and teaches those skills.
First, the basic idea
Start with AI's real capabilities: then compare them with tasks inside a career. The Skills & Careers section helps connect that understanding to personal strengths and learning choices.
Where workers use AI
Developers draft tests, accountants categorise transactions, support teams summarise cases and marketers compare possible messages before a person approves the work.
What AI can help you strengthen
It can reduce repetitive work and reveal alternatives. Your advantage comes from adding field knowledge, checking the result, communicating clearly and taking responsibility.
Artificial Intelligence is transforming every industry. The jobs of the future will require new skills, adaptability, and collaboration with AI systems.
Can a job contain both automatable tasks and tasks that still need human judgement?
Analyse what someone actually does during a working week.
Trace the explanation
Drafting, checking, communicating and taking responsibility can be affected differently. A tool's success on one task does not establish that the whole role can be replaced safely.
Source and further reading: www.ilo.org
A deeper look at Careers in an AI WorldAI Changes Tasks Before It Eliminates Entire Careers
Jobs are bundles of tasks. A teacher plans lessons, explains ideas, marks work, communicates with families and manages a classroom. An accountant interprets rules, checks records, advises clients and prepares reports. AI may automate or accelerate some tasks while leaving others largely human. Looking at the task level gives a clearer picture than asking whether one whole career will “disappear”.
Work that is repetitive, digital and easy to evaluate is usually easier to automate than work requiring physical dexterity in unpredictable environments, deep trust, accountability or complex human judgement. Even then, technology can change the job. A mechanic may use AI-assisted diagnostics; a nurse may use prediction tools; a developer may use code assistants. The worker still needs to understand the result and act responsibly.
This shifts the value of skills. Knowing how to use AI tools is useful, but so are domain knowledge, verification, communication, problem framing and judgement. If everyone can generate a draft in seconds, the person who can recognise whether the draft is correct, relevant and safe becomes more valuable.
Career planning should therefore focus on adaptability instead of trying to find a mythical “AI-proof” occupation. Build a strong base in a field, learn how technology is changing its workflows, practise human skills that are difficult to automate and keep evidence of real work you can do.
Automation usually targets tasks first
A job description may survive while the way the work is performed changes substantially.
New tools create new responsibilities
AI adoption creates demand for evaluation, governance, data quality, integration, security and training.
Domain expertise still matters
An AI tool used in law, medicine, engineering or finance is more useful when the person using it understands the field well enough to spot weak output.
A job contains more tasks than its most automatable example
A designer may generate draft images quickly but still clarify a client's needs, negotiate changes and prepare a usable final product. A developer may generate code while still diagnosing production failures and owning the release. A task demonstration does not measure the whole job.
Map a familiar role into tasks. Mark which have clear inputs, easy-to-check outputs and low consequences if wrong. Then mark tasks involving ambiguous goals, relationships or responsibility. That map is more useful than a list declaring entire careers safe or doomed.
Break a job into tasks to Develop skills across the remaining workflow
A new tool drafts reports quickly. Which evidence would best establish its effect on the whole job?
Trace the explanation
Measure the complete workflow. Faster drafting can be offset by verification or correction work. Also examine tasks the tool does not handle and who remains responsible for the released result.
Source and further reading: www.ilo.org
Which skills can you demonstrate when job titles keep changing?
Evidence of problem-solving transfers better than a claim to know every new tool. A clear work sample shows decisions, corrections and results.
Keep a small project with the original problem, your approach and a checked outcome. Explain where a tool helped and where your judgement changed its suggestion.
Frequently Asked Questions
Questions about careers in an AI world
Will AI replace most careers?
AI tends to take over repetitive tasks instead of whole jobs, so careers that involve judgement, context and accountability stay valuable.
What should I learn to stay relevant as AI improves?
How to direct AI tools effectively, verify their output, and apply the result to a specific real-world situation.
Which industries are changing fastest because of AI?
Customer support, content drafting, data analysis and basic coding are seeing the most visible shifts so far.
Is it too late to prepare for an AI-influenced career?
No. The skills that matter most, learning quickly and checking results, can be built at any career stage.
Think about tasks, not just job titles
AI usually changes parts of a job before it replaces an entire occupation. A job is a bundle of tasks: collecting information, checking details, making judgements, communicating with people, handling exceptions, and taking responsibility. Some tasks are repetitive and easy to describe; others depend on trust, context, physical work, or consequences that require human judgement.
A practical task map
- Good candidates for assistance: summarising a long document, drafting a first version, sorting routine requests, or generating test cases that a person can review.
- Tasks needing careful verification: researching facts, preparing financial explanations, writing code that affects real users, or translating sensitive information.
- Tasks that should not be handed over blindly: decisions about someone's rights, safety, money, education, or employment without accountable human oversight.
For example, a receptionist might use AI to draft a reply to a common question. The receptionist still checks whether the answer matches the organisation's policy, protects private information, and addresses the actual visitor. The useful skill is not simply “prompting”; it is knowing what a good answer should contain and spotting when the tool is wrong.
A six-week learning plan
- Choose a real task from a field that interests you.
- Learn the basic vocabulary and workflow of that field.
- Use an AI tool on a low-risk practice example, without confidential data.
- Check its output against a reliable source or a known correct result.
- Record one error, one limitation, and one improvement you made.
- Build a small portfolio showing the task, your process, the checks, and the final result.
This portfolio gives stronger evidence than claiming to be “good at AI.” It shows that you can define a problem, use a tool sensibly, check quality, and explain your decisions.
What not to automate
Do not automate a process merely because it is repetitive. First ask what happens when the system is wrong, who can challenge its decision, what personal data it receives, and whether a person can recover from a mistake. Keep a manual fallback for high-impact decisions and make responsibilities clear.
Further reading: World Economic Forum, Future of Jobs Report 2025 discusses changing skills and work; NIST AI Risk Management Framework offers a framework for considering risks.
Build evidence that you can work with the tool and judge its output
For a portfolio, show the starting problem, the generated draft, the checks you performed and the corrections you made. Explain where automation saved time and where it introduced risk. A final artefact without the reasoning hides your contribution.
Choose practice that remains useful when a product changes: clear writing, domain knowledge, data interpretation and reliable execution. Learning one interface can help today, but understanding the workflow lets you adapt when buttons and model names change.
If a model performs a task well in a demo, does that prove an entry-level role will vanish?
No. A labour-market outcome also involves adoption costs, demand, workflow changes and accountability. Treat that as a question for evidence, not a conclusion from one output.
A producer uses AI to draft music. What parts of the work might still require deliberate human decisions?
One reasoned response
Consider artistic direction, selection, editing, rights checks, client communication and responsibility for the released work. The workflow changes, but the generated draft alone does not settle those decisions.
Source and further reading: www.ilo.org
Audit a job by its tasks, not its title
AI changes work unevenly. Two people with the same job title may spend their days on very different tasks, so a useful career question is not “Will AI replace this job?” but “Which parts of this work can a tool assist, and which parts still need accountable human judgement?”
Use a four-part task map
- Routine information work: sorting, formatting, summarising or drafting from clear instructions may be easier to assist with.
- Context-heavy decisions: work involving incomplete evidence, competing priorities or local knowledge needs careful human review.
- Physical and interpersonal work: handling unpredictable environments, building trust, caring and negotiating may be difficult to automate end to end.
- Accountability: someone must be responsible for checking outcomes, protecting private information and correcting harm.
A practical example
A customer-support worker might use AI to draft a response to a common question. The worker still checks whether the answer matches the customer's account, whether the policy applies, and whether the situation needs escalation. The tool can reduce repetitive typing; it should not be allowed to invent a refund or expose another customer's details.
Build skills in small steps
Choose one tool relevant to a field you are exploring. Learn its basic operation, practise checking its output, and create a small portfolio example that shows your own decisions. Pair tool familiarity with durable skills: clear writing, numeracy, domain knowledge, problem solving, teamwork and the ability to explain why a result is trustworthy. Avoid training promises that guarantee a job or salary; compare course cost, practice opportunities, entry requirements and evidence of employer demand.
Sources
Sources and further reading
A closer lookChoose a skill you can demonstrate
A job advert asks for spreadsheet analysis, clear communication and AI familiarity. Instead of listing only tool names, an applicant builds a small report and explains how every number was checked.
Work through the distinction
Keep private data out of the project. Show the source, assumptions, corrections and the limits of the conclusion. A portfolio should demonstrate judgement alongside tool use, because a polished output alone does not show how it was produced.
Which evidence is more useful?
Choose an answer to see feedback. You can try again.
Show the answer and explanation
A checked project with explained decisions. Keep private data out of the project. Show the source, assumptions, corrections and the limits of the conclusion. A portfolio should demonstrate judgement alongside tool use, because a polished output alone does not show how it was produced.