Are Graduates Ready for the AI Era? An Honest Look

The old formula still works, but it is no longer enough

For a long time the formula was simple: study hard, get into university, get a degree, land a decent white-collar job. I am not saying that formula has stopped working. Graduates still generally do reasonably well. But the degree on its own is no longer enough.

You can see it in Malaysia already. In the first quarter of 2026, 35.2% of employed people with tertiary education were classified as being in skill-related underemployment. In plain English, more than one in three working graduates are in jobs that do not require their degree. That does not necessarily mean universities failed them. It does tell us that producing more graduates does not automatically produce more high-skill careers. We dig into the numbers in our research on the underemployment paradox.

AI is accelerating the mismatch. A company hiring a fresh graduate does not just care that you studied marketing, computer science or finance. It wants to know whether you can actually do something: solve its problem, communicate, use the tools the team uses, learn fast, and increasingly, use AI well. The labour market is moving much faster than the curriculum, and I do not think the curriculum will ever catch up.

So what were students supposed to do?

I feel for students here. Imagine you are 20. Your lecturer says do not use AI, you need to learn this yourself. Fair enough. Two years later your employer says use AI, because everyone on the team does and they are far more productive. What exactly were you meant to do?

The answer is not banning AI. It is already part of daily life. But the answer is also not typing every assignment into ChatGPT and pasting whatever comes out. Both extremes miss the point. AI is there to sharpen your thinking, not replace it.

We deal with this directly at Sigmaschool. We spent a lot of time on whether students should use AI while learning software engineering, and the answer became obvious: of course they should. That is the environment they will work in. Banning it would be like banning the internet. So they build with AI, debug with AI and research with AI. But at certain points we take the AI away and ask them to explain their decisions, solve a problem, debug something themselves and prove there is a brain behind the prompt. Not an AI-free education, but an AI-native one with real human understanding underneath.

Being AI-ready is not a list of prompts

A common misconception is that being AI-ready means collecting clever prompts. I do not think that ages well. ChatGPT, Claude and the rest are working hard to make the experience effortless, and models keep getting better at understanding what you mean. Prompt engineering will matter less and less for ordinary users.

The real skill is knowing what to ask, recognising whether the answer makes sense, knowing enough about the subject to correct the AI when it goes the wrong way, and turning the output into something useful. That is why domain expertise still matters. A great marketer who understands customers, positioning and distribution becomes far more powerful with AI, like having a pool of geniuses on call. An engineer who understands architecture gets much faster. A lawyer who knows the law can analyse information dramatically faster.

But if you know nothing about your field, AI does not make you an expert. It may actually make things more dangerous as models improve, because bad output sounds very convincing. The future does not belong to people who learn AI instead of a profession. It belongs to people who know their profession and use AI to multiply themselves.

Human skills multiplied by AI skills

The World Economic Forum surveyed employers and found they expect around 39% of workers' core skills to change by 2030. Not 39% of jobs disappearing: 39% of the skills people use in their jobs today changing. AI and big data are among the fastest-growing skill areas. But employers are also placing more weight on analytical thinking, creativity, resilience and lifelong learning.

So it is not human skills versus AI skills. It is human skills multiplied by AI skills. If your human skills are zero, multiplying by AI still gives you zero. Someone who thinks clearly, communicates well, understands their domain, learns quickly and uses AI effectively becomes very valuable, very fast.

This is actually the opening for juniors. There is a real gap between seniors who have not learned to use AI and newcomers who have solid fundamentals and know how to multiply them with AI. That is your chance to stand out. We wrote more about it in the AI-native junior developer.

Universities and employers need to be closer

Right now there is an awkward gap. University teaches one set of things, you start work, and the company wonders why you cannot do basic things it needs and starts teaching you from scratch. Why are those two worlds not connected?

I think we will see far more apprenticeships, internships, industry projects, bootcamps, shorter qualifications and micro-credentials sitting alongside degrees. Imagine studying computer science but spending part of it building software with real companies, or studying finance while doing placements inside financial firms. That is a much stronger signal to an employer, because you do not just have education. You have evidence.

The argument against university goes too far

I want to be fair here. Plenty of professions absolutely require university: medicine, law, engineering and many specialised, regulated fields that need deep foundations. Even outside those, university gives you structured learning, a network, independence, exposure to different people and fields, and several years to work out what you are actually interested in.

There is another reason deep education matters in an AI world: to judge AI properly, you need knowledge. Someone who understands a subject deeply can tell when AI is producing nonsense. Someone who knows nothing just accepts it. Ironically, AI does not make expertise less important. It makes genuine expertise more valuable.

What is disappearing is the idea that the degree itself is enough. A prestigious piece of paper is not an automatic career any more. If you are wrestling with this, read is my degree worthless? for a longer take.

What I would do if I were 18 today

I would not try to predict the exact job that AI will leave alone for the next 40 years. Nobody knows. Anyone who claims to is guessing. Instead:

Get comfortable being a beginner again

Finally, I would make myself comfortable being a beginner, over and over. With nearly 40% of core workplace skills expected to change within a few years, the person who wins is not the one who knows the most today. It is the one who can learn and adapt fastest tomorrow.

So, are graduates prepared for the AI era? Some are. I do not think the education system as a whole has caught up, and if you are a student today you cannot wait for it to. A degree can still be valuable. Domain expertise still matters. Critical thinking matters most of all. None of that removes the need for AI literacy, and AI literacy does not remove the need to actually understand something. You need both.

The biggest career risk over the next few years probably is not AI suddenly taking your job. It is someone in your field learning to work with it much better than you. If you want to build both halves at once, software is one of the best places to do it, because you learn real fundamentals and real AI skills together. Our free trial is a no-commitment way to start, and the full AI-Native Software Development Programme is there when you are ready.