No Entry: How AI is Closing the Door on Graduate Work Before Many of Us Can Get In

Australian graduate full-time employment fell sharply in 2024, and the drop was not random. 
The fields most exposed to AI disruption saw the steepest declines. 
For international students who came here to build a career in exactly those fields, the numbers tell a story that is hard to ignore.

Data visualisation story | Topic 1: AI and Humanity | Format: Five Charts
Data sources: QILT GOS 2024, QILT GOS-L 2025, ABS Labour Force, OECD (2026), WEF Future of Jobs 2025

I spent the better part of four years in rooms where people talked about AI in the future tense. This was Gurugram, 2022 to mid 2025, and I was working as a business analyst at Cognizant, stationed at MG Motor India as their single point of contact for SAP Aftersales and Sales technical support. I worked with their internal teams, dealership networks, and business stakeholders, built Excel automations, produced dashboards in Power BI, and sat through meetings where AI kept coming up as something the bigger teams were working towards. It always felt like it mattered. It never actually arrived, at least not where I was sitting.

So I came to Melbourne. That is the honest version of why I enrolled in a Master of Data Science at RMIT. I could see from Delhi that the technical floor was rising fast, and I wanted to build the kind of depth that keeps you above it rather than under it. A Masters felt like the right move: strategic, deliberate, and grounded in nearly four years of real business experience that I could bring into a more technical role once I graduated.

What I did not anticipate was that the floor would rise so fast specifically at the entry point of the market, and that this would happen while I was still mid-degree and actively looking.

Right now I am in my first year at RMIT with two semesters ahead of me. On weekends I work at T2 Tea in Fitzroy and run the floor at Criniti’s, an Italian restaurant in Southbank, which is how I cover rent while I build a portfolio and apply for the part-time data analyst and business analyst roles that are supposed to bridge studying and working in this field. Those applications have been going nowhere. Not rejections. Just silence. Roles appear on LinkedIn one week and disappear the next. I started asking whether this was a personal problem or a structural one.

It is structural. Here is what the data shows.

The 2024 drop reversed two years of post-pandemic recovery in graduate employment

The graduate employment data is collected by the Quality Indicators for Learning and Teaching (QILT) program, which surveys graduates four to six months after they finish their degree. It is one of the most reliable snapshots of what actually happens to Australian university graduates when they enter the labour market.

What that data shows for 2024 is striking. Domestic undergraduate full-time employment fell from 79.0 per cent in 2023 to 74.0 per cent in 2024, the largest single-year drop in the survey’s history outside the pandemic year of 2020. Postgraduate coursework employment also declined, from 90.3 per cent to 88.1 per cent, though the fall was smaller because postgraduate graduates tend to enter more senior roles with more specific skill requirements.

It would be easy to attribute this entirely to broader economic conditions: interest rates, a cooling labour market, and employer caution. And those factors are real. But that explanation does not tell us why the drop was so much larger in some fields than others. For that, you need to look at where the AI exposure is highest.

Fields with the highest AI exposure saw the steepest employment drops in 2024

This chart makes the structural story visible. Fields sitting in the top-left quadrant, where AI exposure is low and employment is improving or stable, are mostly protected professions: nursing, medicine, teacher education, engineering. Fields in the bottom-right, where AI exposure is high and employment has dropped, are concentrated in the areas where AI has already demonstrated competence: writing, data processing, administrative reasoning, basic coding, and structured communication.

Computing and information systems dropped 6.6 percentage points. Business and management dropped 6.0. Communications dropped 6.6. These are not small movements in a survey this size. They are signals.

There is a legitimate debate about causation. Andrew Norton’s September 2025 analysis of the same data notes that the Reserve Bank’s aggressive rate-hiking cycle in 2022 and 2023 may have triggered a hiring freeze that preceded AI adoption at scale, and that it is hard to cleanly separate AI’s effect from broader economic tightening. That is fair. But the field-by-field variation, the fact that nursing and medicine barely moved while computing and communications fell sharply, is harder to explain through interest rates alone. Rate hikes do not discriminate between a nurse and a data analyst when it comes to hiring appetite.

The tasks graduates spend their first years doing are the same ones AI now does well

The OECD’s 2026 AI Capability Gap Index, published just weeks ago, offers the most rigorous available measure of where current AI systems sit relative to what different jobs actually require. The index rates AI capability across nine domains: language, social interaction, problem-solving, creativity, metacognition, knowledge and memory, vision, manipulation, and robotic intelligence.

The findings that matter most for this story are in the language, knowledge, and administrative domains, where the gap between AI capability and human performance has effectively closed for routine tasks. Data entry, document drafting, scheduling, research synthesis from structured sources: these are the tasks that dominate the first one or two years of a graduate’s working life. They are how new workers earn trust, demonstrate capability, and get promoted into more complex work. That ladder is now being automated away at the bottom rung.

The concern is not that AI replaces the entire role of a data analyst or communications graduate. It is that AI replaces the specific slice of work that justifies hiring someone at entry level in the first place. If a senior employee can do three times the work with AI tools, the junior headcount that used to support them becomes hard to justify in the next budget cycle.

International graduates in high-AI-exposure fields face a wider employment gap than domestic peers

This chart makes the personal story structural. The international versus domestic employment gap is not new. It has existed for years, driven by factors including visa conditions that restrict work rights, employer preferences for domestic networks, and the extra friction of professional migration. But the gap is widest in computing and business: 25.6 percentage points and 25.5 percentage points respectively at the short-term mark.

Those are the two fields where AI exposure is highest. They are also the two fields that international students from countries like India most commonly enrol in. Students who came here specifically to build skills in the technology and business sectors are entering a narrowing market, carrying a double disadvantage: the AI disruption hitting their target fields, and the structural disadvantage that has always existed for international graduates trying to break in.

There is some comfort in the medium-term numbers. Three years after graduation, international graduates in engineering reach 90.6 per cent full-time employment, almost matching domestic peers. Science and mathematics reach 79.7 per cent. The gap does close, eventually. But the short-term picture is the one that matters for someone mid-degree and actively building towards their first role in the field. The gap does not feel abstract when you are living inside it.

The roles expanding fastest by 2030 are ones where human judgment still matters most

The story does not end here. The World Economic Forum’s Future of Jobs Report 2025, drawing on survey data from over one thousand employers across fifty-five economies, projects that 170 million new roles will be created globally by 2030, even as 92 million existing ones are displaced. The net figure is positive, but the distribution matters enormously: the roles being created require a fundamentally different skill profile from the ones being automated away.

What the fastest-growing roles share is a consistent need for something AI cannot yet replicate at scale: contextual human judgment, trust, accountability, and physical or social presence. AI and machine learning specialists need to understand when a model is wrong and why. Cybersecurity professionals need to anticipate adversarial human behaviour. Environmental scientists need to navigate complex ecological and political uncertainty. Healthcare practitioners need to sit with patients in the room.

For graduates in computing and data science, and this is the field I am in, this means the career path is shifting from learning the tools to learning when and why to use the tools, and how to take responsibility for what they produce. That is a more demanding credential, but it is also a more durable one. The students who build portfolios demonstrating that kind of judgment, rather than just demonstrating technical familiarity, are the ones who will find the door stays open.

Which is exactly what I am working towards, building the kind of portfolio that demonstrates judgement rather than just technical familiarity, one project at a time.

The data presented in these five charts does not tell a simple story of doom. It tells a story of structural disruption that is landing unevenly. It is hitting hardest at the bottom of the labour market ladder, in the most AI-exposed fields, for the most vulnerable workers. For international students in their mid-to-late twenties, who moved countries for a qualification in exactly those fields and are now building portfolios in their spare time between hospitality shifts, that structural landing is very personal.

The right policy response is not to pretend the disruption is not happening. It is to design graduate employment programmes, international student work rights, and university curriculum that acknowledges where AI has genuinely changed what entry-level means. The ladder has not disappeared. Its bottom rung is just higher than it used to be.