For decades, graduate jobs followed a simple pattern. New starters did the routine work: first drafts, data entry, research summaries, basic analysis. In return, they learned the job from the ground up.
AI now does a lot of that routine work, and does it fast.
That leaves employers with a real question. If the junior tasks are disappearing, what are graduates for? And how do you choose the right ones when every application looks polished?
What is happening to graduate hiring
The numbers show a clear squeeze. The Institute of Student Employers reported that graduate hiring fell by 8% in its 2025 survey, the weakest year since the pandemic, with a further fall forecast. At the same time, the average employer received around 140 applications for every graduate vacancy, the highest level the ISE has recorded.
Part of that surge comes from AI itself. Candidates can now use AI tools and "easy apply" buttons to send dozens of tailored applications in an afternoon. Employers get more applications, but less signal from each one.
So employers face two problems at once:
- Fewer traditional junior tasks to build a graduate role around.
- More applications than ever, and harder to tell apart.
Why cutting graduate hiring is a mistake
It is tempting to stop hiring graduates altogether and let AI cover the gap. Some businesses already have.
The risk is that it creates a hole in your future. Today's graduates are tomorrow's managers, specialists and leaders. If nobody joins at the bottom, there is nobody to promote in five years' time. You end up paying more to hire experienced people from outside, if you can find them at all.
AI also needs people who can check its work. Someone has to spot when an AI summary is wrong, when a forecast does not make sense or when a draft misses the point. That takes judgement, and judgement has to be learned somewhere.
The smart move is not to hire fewer graduates. It is to hire graduates for different reasons.
What graduate roles look like now
When AI handles the routine work, the human part of a junior role shifts towards:
- Checking and improving AI output, not just producing first drafts.
- Working with people: customers, colleagues and suppliers.
- Solving problems that don't have a template.
- Learning fast, as tools and processes keep changing.
- Making sensible decisions with incomplete information.
These are not new skills. But they used to be things graduates grew into over a few years. Now they are needed much earlier.
What to test for in graduates
A degree tells you that someone can study. It does not tell you how they think, how they work with others or how quickly they will learn. And a CV written with AI tells you even less.
Here is what is worth measuring instead.
Reasoning and learning ability
Cognitive ability is one of the best general signs of how quickly someone will pick up a new role. Numerical, verbal and inductive reasoning tests show whether a candidate can work with data, read carefully and spot patterns, which is exactly what you need to check AI output.
Critical checking
Someone who will review AI work needs a sharp eye for mistakes. An error checking test shows how carefully a candidate works and whether they catch what others miss.
Judgement
Situational judgement tests put candidates in realistic work situations and ask what they would do. They are one of the best ways to see how a graduate with little work experience is likely to behave on the job.
Adaptability and problem solving
When the job keeps changing, the ability to adjust matters. Adaptability and problem solving tests help you find people who are comfortable with change and who can work out what to do when there is no clear process.
Communication and teamwork
As routine tasks move to AI, more of a graduate's day involves people. Communication and teamwork skills are now central, not extras.
Working style
A personality test can show how someone prefers to work, what motivates them and where they are likely to thrive. For graduates with short CVs, this fills in much of the picture.
How to build a better graduate process
1. Test first, read CVs second. With so many applications, an early online assessment is the fairest and fastest way to build a shortlist. Every candidate gets the same chance to show what they can do. For a ready-made starting point, see our graduate trainee assessment.
2. Look beyond the degree. Some of the best candidates may not come from the universities you usually hire from, or from university at all. Testing for ability rather than background widens your pool. Our apprentice and intern assessments work the same way.
3. Use timed, randomised tests. Graduates are confident AI users. Timed tests with a different set of questions for each person make sure the result reflects the candidate, not a chatbot. (We cover this in more depth in Can candidates use AI to pass your tests?)
4. Redesign the role, then hire for it. Before you advertise, decide what your graduates will actually do now that AI covers the routine work. Then test for those skills.
5. Train them to work with AI. Give graduates real responsibility for checking, improving and explaining AI output. It is the fastest way to build the judgement your business will need later.
The bottom line
AI is changing what graduate jobs look like, but it has not removed the need for them. If anything, the graduates you hire now need to be sharper, more adaptable and better with people than ever before.
The CV cannot tell you who those people are. Objective testing can.
Test Candidates helps employers look past polished applications and see the real ability, judgement and working style of every graduate applicant. See how it works or start free.
Frequently asked questions
Is AI replacing graduate jobs?
AI is replacing many of the routine tasks that graduates used to do, and some employers have reduced graduate hiring. But the need for new talent has not gone away. The roles are changing towards checking AI work, solving problems and working with people.
Why are there so many applications for graduate jobs?
Fewer vacancies and AI tools that make applying faster have pushed applications per vacancy to record levels. That makes it harder to stand out, and harder for employers to choose.
What is the best way to assess graduates?
Use a short online assessment early in the process that covers reasoning, judgement and soft skills. These show potential more clearly than a degree or CV, especially for people with little work experience.
Should graduates be allowed to use AI at work?
Yes, in most roles. Knowing how to use AI well is a valuable skill. The key is making sure they also have the reasoning and judgement to spot when AI gets it wrong.
