For generations, the transition from education to employment followed a reasonably predictable sequence.

Schools and universities provided foundational knowledge. Employers hired young people with limited experience into junior positions. Those employees then learned how work actually happened how to communicate professionally, interpret instructions, analyse information, deal with clients, check their work and exercise judgment.

Artificial intelligence is beginning to disturb this arrangement.

The concern is not simply that AI could replace jobs. A more immediate change is happening inside jobs: AI can increasingly perform some of the routine tasks traditionally assigned to people at the beginning of their careers.

That creates an unusual problem for education:

If AI performs some of the work through which beginners used to become experienced workers, where will tomorrow’s workers acquire that experience?

AI may be changing the first rung of the career ladder

The International Labour Organization’s 2025 assessment estimates that approximately 24% of global employment around 838 million jobs is in occupations with some exposure to generative AI.

This does not mean that one-quarter of jobs will disappear. The more useful interpretation is that particular tasks within occupations can increasingly be performed or assisted by AI. The ILO therefore considers job transformation more likely than wholesale job replacement.

But where that exposure occurs matters.

Among the occupations with substantial exposure are clerical and administrative activities such as data entry, bookkeeping and information processing. These are precisely the kinds of tasks that have traditionally provided young workers with their first exposure to professional work.

A junior employee may begin by preparing basic reports, conducting preliminary research, responding to routine enquiries or checking documents. The task itself may not require sophisticated judgment, but repeatedly doing it builds domain vocabulary, attention to detail, professional discipline and an understanding of how an organisation works.

If AI performs more of this foundational work, companies may eventually require entry-level employees to contribute at a higher level much earlier.

The danger, therefore, is not necessarily the disappearance of the entire entry-level job.

It is the disappearance of some of the entry-level learning process.

Employers want technology and human capability

One response might be to teach every student more technology.

The labour-market evidence suggests something more complicated.

The World Economic Forum’s Future of Jobs Report 2025 identifies AI and big data as the fastest-rising skill area towards 2030, followed by networks and cybersecurity and technological literacy.

Yet the capabilities employers consider most important today remain strongly human and cognitive.

The message is more useful than simply saying that students need “AI skills.”

Employers are not choosing:

Technology OR human capability.

They increasingly need:

Technology + analytical ability + judgment + adaptability + communication + creativity.

Knowing how to use an AI tool may therefore soon be no more differentiating than knowing how to use a search engine or spreadsheet. The valuable capability becomes knowing when to use it, what to ask it, whether its answer is credible, what it has missed and when human judgment should override it.

Using AI is not the same as learning with AI

Schools face another complication: giving students access to AI does not automatically make them better prepared.

OECD’s PISA 2025 examined generative-AI use among 15-year-olds across participating education systems. The findings suggest that outcomes depend substantially on how AI is used and whether students are taught to evaluate it critically.

Students using AI for particular schoolwork activities did not automatically perform better. The evidence also indicates that combining AI use with school-based AI-literacy instruction produces different outcomes from simply allowing students to use the technology.

This distinction matters.

An education system can produce students who are frequent AI users without producing students who are AI literate.

AI literacy therefore needs to go beyond prompting. Students should be able to ask:

Where did this answer come from? Is the evidence credible? What assumptions has AI made? What information is missing? Could another interpretation be possible? Would I recognise if the answer were wrong?

Ironically, AI may therefore make foundational knowledge more, not less, important. A student cannot critically evaluate an AI-generated economic argument, scientific explanation or calculation without understanding enough of the subject to recognise errors.

Schools should consequently avoid two extremes:

Ignoring AI and allowing AI to substitute for learning.

Neither prepares students for an AI-enabled workplace.

India’s challenge is larger than adding AI to the curriculum

India already has a policy foundation for more work-oriented education.

The National Education Policy 2020 and National Curriculum Framework 2023 envisage greater exposure to vocational education, practical work and transferable skills.

Implementation, however, remains a challenge. The underlying data show that 18,610 secondary and higher-secondary schools offered NSQF vocational courses in 2023–24, against approximately 2.88 lakh secondary and higher-secondary schools nationally, an implied coverage of only around 6.5%.

Employer evidence raises another concern.

The World Bank’s Jobs at Your Doorstep work across six Indian states found recurring demand across very different sectors for problem-solving, teamwork, adaptability and the ability to use digital technologies. It also identified mismatches between vocational trades offered by schools and employment demand in surrounding districts.

Simply increasing the number of vocational subjects will therefore not solve the problem.

The relevant measure needs to shift from:

“Does the school offer vocational education?”

to:

“Can the student perform useful work in a changing workplace?”

Schools may need to become part of the first job

This is where AI requires a deeper rethink.

Traditionally:

Education → Entry-level job → Workplace learning → Experienced professional

If basic entry-level tasks become increasingly automated, part of that learning may have to move forward:

Education + applied workplace experience → AI-enabled entry-level job → Higher-value workplace learning

Schools do not need to become companies. But students need more opportunities to deal with the type of ambiguity that characterises actual work.

A business student could analyse a real small-business problem. A science student could interpret imperfect experimental data. A humanities student could investigate conflicting sources and defend an evidence-based conclusion.

Students could also receive AI-generated outputs containing weaknesses and be required to detect errors, verify evidence and improve the analysis.

Assessment would then test not merely what students can produce, but how they think when technology can produce the first draft for them.

What should change?

The evidence points towards four practical shifts.

1. From using AI → to supervising AI

AI literacy should include verification, source evaluation, bias recognition, reasoning, privacy and accountability. Prompting is useful; prompting alone is not literacy.

2. From learning first and working later → to earlier workplace exposure

Internships, employer projects, apprenticeships, applied capstones and realistic simulations can provide some of the experience previously accumulated through junior work.

3. From static vocational courses → to transferable capabilities

Technical skills matter, but students also need problem-solving, communication, teamwork, digital fluency and adaptability. Vocational offerings should additionally reflect actual labour-market opportunities around students.

4. From assessing output → to assessing reasoning

If AI can produce a competent essay, summary or presentation, the assessment question needs to change.

Instead of only:

“What did you produce?”

schools can increasingly ask:

“Why did you reach this conclusion? What evidence supports it? How did you use AI? What did you reject? What would make you change your answer?”

Can schools still prepare students for work?

Schools will probably never update curricula as quickly as technology changes.

They do not need to.

Their more durable responsibility is to prepare students for a labour market in which tools, tasks and even occupations will repeatedly change.

The old education-to-employment contract was relatively simple:

School teaches the foundations → the first job teaches you how to work.

AI is beginning to blur that division.

If entry-level employees are increasingly expected to arrive capable of working with AI, checking its output, solving less-structured problems and exercising judgment, then some learning previously left to the first job must begin before the first job.

So, can schools still prepare students for work when AI is changing entry-level jobs?

Yes, but only if preparing students for work stops meaning preparing them to perform yesterday’s entry-level tasks.

The goal should not be to predict every technology students will encounter. It should be to ensure that when the technology changes, they can still understand the problem, learn the tool, question its output, work with people and take responsibility for the final decision.