Blog · The Shift
Which Skills AI Makes Obsolete, and Which It Rewards
6 November 2026 · 8 min read
The skills AI makes obsolete share one profile: the work is repetitive, well documented and easy to check. The skills it rewards share the opposite profile: they need judgement about a specific business, context that is not written down anywhere, and someone willing to answer for the result. Build your career on the second list, and use AI for the first.
This guide gives you a simple test you can apply to any skill you have, two lists to compare it against, and a practical way to move your time from one list to the other. It is written for working professionals, aspiring founders and SME owners in India, the UAE, Kenya and online who want a clear view rather than another warning.
A Simple Test for Any Skill
Ask one question: could a capable AI tool do most of this skill, given clear instructions and a few good examples? If the honest answer is yes, the market value of that skill is falling, however hard it was to learn and however long it took you.
That last part is uncomfortable. Many people spent years becoming fast and accurate at work that is now cheap. The years were not wasted, because the understanding they built still matters, but the speed itself no longer sets you apart.
Three Questions That Sharpen the Test
- Is the task repetitive? Does it follow the same pattern most of the time?
- Is it well documented? Is there a large amount of public material showing how it is done? AI tools learn from exactly that material.
- Is the result easy to check? Can someone tell quickly whether the output is right?
Three yeses describe the skills AI makes obsolete fastest. Three noes describe skills that are likely to hold their value for longer.
The Skills AI Makes Obsolete Fastest
Skills under pressure share the profile above. Common examples include:
- First drafts of routine documents, emails and reports.
- Routine translation and summarising.
- Data entry, clean-up and reformatting.
- Standard reports built from the same template each month.
- Template design and simple visual layouts.
- Boilerplate code for common tasks.
- Scripted first-line support.
The research on exposure matches this list. The International Labour Organization's 2025 index of occupational exposure to generative AI found that clerical occupations remain the most exposed, while concluding that transformation of jobs is a more likely impact than full automation (ILO, 2025). In plain terms: the skills AI makes obsolete tend to be tasks within jobs, and roles built almost entirely on them feel the change first.
Obsolete Does Not Mean Useless
A skill losing market value is not the same as a skill losing all value. Knowing how a good report is structured still helps you judge an AI-drafted one. Knowing how data entry goes wrong helps you design a check. The difference is that the skill becomes an input to judgement rather than a job in itself.
Which Skills Does AI Reward?
The skills AI rewards share the opposite profile. They depend on context about a specific business that is not written down, they are hard to check quickly, and someone has to take responsibility for them.
- Choosing the problem: deciding which problem in a business is worth solving at all.
- Scoping a process: drawing where work starts and ends, and where it waits.
- Setting the rules: writing the instructions and limits an AI system works within.
- Checking against reality: testing AI output against what actually happens, not what looks plausible.
- Owning the decision: taking responsibility for the result, including when it goes wrong.
Why Checking Output Is Now a Core Skill
The checking skill deserves special attention because even heavy users of AI do not fully trust it. In the Stack Overflow Developer Survey 2025, 84% of respondents said they use or plan to use AI tools, yet 46% said they distrust the accuracy of AI output against 33% who trust it, and the most cited frustration (66%) was AI solutions that are almost right, but not quite (Stack Overflow, 2025). That gap between "almost right" and "right" is exactly where a person with judgement earns their place.
The Second Group: People Who Combine Tools
A second group also gains value: people who can combine tools. A person who can connect an AI model to a spreadsheet, a customer database and a payment link, and make the whole thing run one process, is doing what used to need a small team. This is not programming in the traditional sense. It is closer to designing a workflow and then assembling it from parts, which we cover in building a working tool without writing code.
How people already use AI suggests this is where things are heading. Anthropic's Economic Index (February 2025) found that use leaned more towards augmentation (57%) than automation (43%), meaning people mostly use AI to extend their own work rather than hand it over entirely (Anthropic, 2025). Combining tools well is the skill that turns augmentation into a working process.
How to Move Your Time From One List to the Other
The practical move is not to compete with AI on the first list. It is to use AI for the first list and spend your time on the second. A simple way to start:
- Audit a fortnight of your work. List every task and mark it against the three questions above.
- Hand one routine task to a tool. Choose something low risk, run it with AI for two weeks, and check every output.
- Write down where it failed. Those failures show you where judgement is still needed, and how to judge an AI tool against a real task properly.
- Use the time saved on a judgement task. Map a process, cost a change or design a check.
- Repeat with a bigger task. Each round moves more of your week onto the list that holds its value.
That shift is learnable in weeks, not years, if you practise it on real businesses rather than in theory.
An Illustrative Example: One Role, Two Lists
Take a hypothetical operations executive at a mid-sized distributor. A typical week includes compiling a sales report from three spreadsheets, answering routine customer queries about order status, drafting supplier emails, checking invoices against purchase orders, and sitting in a weekly meeting where late deliveries are discussed.
Run the three questions over that week. The report, the status replies and the supplier emails are repetitive, well documented and easy to check: they sit on the first list. Invoice matching is partly routine, but deciding what to do about a mismatch with a long-standing supplier is not. The late-delivery discussion needs knowledge of which customers matter most and which suppliers can be pushed: that is firmly on the second list.
Moved deliberately, the same person could hand the report and the routine replies to tools, check them, and spend the freed time mapping why deliveries run late and what each delay costs. The role does not disappear. It changes from producing documents to running and improving a process. This example is illustrative, not a case result; the point is the method, which you can apply to your own week.
Does This Look Different in India, the UAE or Kenya?
The test is the same everywhere, but the mix of work differs. In India, large IT and back-office sectors mean many professionals hold routine-heavy roles, along with deep process knowledge that transfers well to the second list. In the UAE, fast-adopting firms often look for people who can run lean teams with AI. In Kenya and across Africa, many businesses are young enough to design processes around AI from the start, which rewards people who can combine tools and set rules early. Our pieces on what the Gulf wants from applied AI and on the broader AI layoffs pattern go further on these differences.
Learning the Second List in Practice
The AI Entrepreneurship Programme is built around the skills on the second list: judging tools, mapping processes, building without code, delegating safely to agents and costing the change. It is twelve weeks of business consulting and training, eight weeks guided and four weeks of remote practical work, in Navi Mumbai, Dubai, Abu Dhabi, Nairobi and online. You can see the week-by-week sequence in what you learn, or use Pathfinder to see how it fits your field.
The programme is training. It does not promise a job, a placement or any business outcome.
Sources
Frequently Asked Questions
What makes a skill vulnerable to AI?
Three features together: the work is repetitive, it is well documented in public material that AI tools learn from, and its result is easy to check. The more of these a skill has, the faster its market value falls.
Is writing a skill AI makes obsolete?
Routine first drafts are under pressure. Writing that depends on knowing a specific audience, business or argument, and on judging what to leave out, is still valuable, and it is needed to edit AI drafts well.
Is data analysis still a valuable skill?
Routine analysis, such as building the same report each month, is increasingly automated. Deciding which question to ask of the data, and spotting when a result does not match reality, is more valuable than before.
How long does it take to move to the skills AI rewards?
The basic shift can be practised in weeks if you work on real processes. Becoming genuinely good at judging, scoping and costing takes longer and comes from repeated practice on different businesses.
Do I need technical skills to combine AI tools?
Not programming. You need to understand the process you are changing, specify clearly what each step must do, and test the result honestly. Many current tools let you connect steps without writing code.
Why does checking AI output matter so much?
Because AI output is often almost right. In the Stack Overflow 2025 survey, two thirds of developers named almost-right answers as their top frustration. Someone has to catch the difference before it reaches a customer.
Should I stop doing tasks AI can do?
Not entirely. Use AI to do them, but keep enough hands-on understanding to judge the output. The aim is to spend less of your time on the routine version and more on the judgement around it.
Will the AI Entrepreneurship Programme change my career?
No. It is business consulting and training in applied AI and does not guarantee a job, a career change or any outcome. What you take from it depends on the work you put in.
Pathfinder · Free
Where Do You Fit in the AI Economy?
Type your industry and the work you do. See the live businesses you would learn inside, your twelve weeks, and three venture ideas made for you.
Find My Path
Join the Discussion
Agree, disagree or add your own experience. Every comment is reviewed before it appears; spam and abuse are removed, and disagreement is welcome.
No comments yet. Be the first to share a view.