As AI agents take on more coding work, IT graduates stay valuable through strong fundamentals, sound judgement, and the responsibility to verify what AI produces.
If AI agents can write code and build whole applications without human intervention, why do I need to study Information Technology (IT)? Could AI replace IT graduates altogether? This is a common question facing current and prospective IT students, and they are fully understandable.
Looking at the current pace of AI development, with thousands of AI agents working in parallel to solve famous maths problems, it looks like we can simply leave building IT solutions in the hands of AI agents. Many also expect that as AI agents continue to improve, they will automate an increasing amount of entry-level work, changing the work that fresh graduates carry out.
So, will AI replace IT graduates? Perhaps a more important question to ask is: what makes someone valuable when AI can do more?
Agentic AI as a Force Multiplier
Unlike early Large Language Models (LLMs), which functioned as chatbots that simply answered questions, an AI agent can use tools to carry out multiple steps towards a goal, such as building and testing software. In this way, it serves as a force multiplier, allowing a person to accomplish more with the same time and resources. For example, a developer might use AI agents to investigate unfamiliar code, compare approaches, and produce a solution more quickly.
Why judgement still matters
As implementations become easier, attention shifts towards deciding what to build, evaluating alternatives, and checking whether the software solution meets real users’ needs. Yet this shift towards higher-value work does not happen automatically, because it requires knowledge and judgement that developers build when they study Information Technology. In other words, delegating to AI agents is useful only if we can recognise whether the agent has done the task well.
Trust Begins with Responsibility
Employers need more than someone who can simply write code; they also need confidence that the resulting system will work reliably and protect its users. Trust rests on responsibility, which in IT involves understanding the requirements, recognising and managing risks, checking results, and responding when something goes wrong. Using AI does not remove these obligations from the people and organisations deploying the software.
For instance, imagine that an AI agent builds a registration website for a student club. The pages look polished and the registration form works, but can one member view another member’s personal details? Someone must decide who should have access, verify that access matches the chosen restrictions, and make sure the team can detect and correct problems. These questions and decisions become even more consequential when systems handle financial records or customer information, and that accountability is one reason organisations cannot simply let AI replace IT graduates.
Powerful Tools Still Need Safeguards
Despite their powerful capabilities, AI agents are not perfect. They can produce incorrect answers, reflect biases in their training, and respond differently to similar prompts. They are also susceptible to prompt injections, where attackers hide malicious instructions in the material an agent reads in an attempt to redirect its actions. As with cybersecurity, managing these risks requires continuous attention as systems and attacks evolve. As a result, we cannot accept an agent’s output simply because it looks convincing. Reviewing and testing the result is part of our responsibilities when using AI, not something optional that we can choose to do.
Use AI to Strengthen Your Fundamentals
Wil AI replace IT graduates who know their fundamentals?
For students who study Information Technology, programming, databases, networking, security and software testing are not merely hurdles to completing a degree. Instead, these subjects provide the knowledge needed to question an agent’s choices and investigate its mistakes. Communication is also essential. Just because a solution is technically impressive doesn’t necessarily mean that it is effective; it can still be solving the wrong problem. Expertise also means recognising when a decision requires specialised advice rather than trusting either yourself or an AI agent without sufficient background knowledge.
Making AI a learning partner
Treat AI as a learning partner rather than a shortcut around learning. Students who study Information Technology should attempt problems before looking at the complete solution, then use AI to explore alternative explanations or challenge their thinking. Used this way, AI can guide the learning experience and develop higher-order thinking skills. Learning works best when students take an investigative approach. That means using AI to dive deep into the topic by testing the proposed answer and considering how it might fail, then practising how to explain the concepts without AI’s help. After all, although AI can make it easy to produce answers, understanding why they are correct remains essential.
Learn to Steer, Not Simply Follow
As AI agents improve, choosing to study Information Technology and prepare for a career in IT means learning how to use increasingly capable tools while continuing to develop independent judgement. Your contributions go beyond producing outputs and extend to the ability to explain, verify, and take responsibility for the work within your role.
So instead of asking ‘will AI replace IT graduates?”, revisit the fundamentals, remain curious and use AI to extend your abilities instead of replacing your understanding. When facing an unfamiliar problem, use AI to understand the underlying principles, then apply those principles to solve the problem. Ultimately, if we are not learning to steer AI, we risk being steered by it instead.
The opinions expressed in this article are the author’s own and do not reflect the views of Swinburne University of Technology Sarawak Campus. Jason Thomas Chew is a Lecturer with the Faculty of Engineering, Computing and Science at Swinburne University of Technology Sarawak Campus He can be contacted at [email protected]