It’s not just students who have quickly embraced generative AI. However, in academia, this poses potential challenges. Should education authorities’ responses focus on how to limit students’ use of AI, or should they focus on educating students on how to use AI responsibly?
Across the UK, the answers to this question vary. Early findings from the Department for Education’s AI Early Adopter Program show that schools and universities are already experimenting with a range of approaches, from cautious restrictions to aggressive encouragement. This is normal for new technologies that are widely adopted. After all, many educational institutions are still working on mobile phones.
However, as a result of this patchwork, different adoption patterns are emerging among students. Some students are learning to use AI tools productively and responsibly; Some people risk completely outsourcing their thinking to themselves. Some people don’t use them all. This is important because it leads to educational inequity.
In this context, the further education sector could play an important role in defining better approaches. It sits at the intersection of academic learning, technical skills and direct input from employers, and is well placed to lead the development of a broader definition of constructive AI use.
It means moving beyond unhelpful outright bans and vague notions of “responsible use” to clear standards and forms of assessment that make student outcomes reliable and comparable. If it can be established here, it will set a precedent for other parts of education, as well as how educational institutions and employers recognize and trust AI-enhanced skills.
AI capabilities are becoming a core criterion for employability
Employers across many sectors increasingly expect new employees to work confidently with AI, so it’s important to use AI appropriately, as well as improve student learning. AI fluency is becoming what digital literacy was a decade ago, becoming a basic expectation rather than a specialized skill. In practice, this looks like employers expecting to delegate certain tasks to AI, frustrated by time-consuming, repetitive manual workflows, and expecting a corresponding increase in productivity.
The AI literacy required for this changed landscape consists of two layers. One is not just knowing how to encourage and get the most out of AI as a tool, but also making decisions about when to use it (or not). The first one can be learned quickly, but as in other fields, judgment can only be gained through long-term learning experiences.
This creates tension for further education providers. They need to maintain confidence in how they assess students while preparing them for a workplace where the use of AI is commonplace. Without the ability to question output, explain reasoning, and validate results, learners risk entering the workforce with skills that do not match expectations.
A better approach is to shift the focus from whether students should use AI to how they can use it in ways that support learning. That means setting expectations, teaching judgment, and aligning assessments with real-world practice. This could strengthen both learning outcomes and the credibility of the qualification.
Prohibition creates inequality, structured guidance creates equity.
Another unavoidable problem with a blanket ban as the default approach to AI is that it does not preclude its use. Instead, it forces it into an unstructured and unsupervised context, leaving learners to interact with various and powerful tools without guidance. As a result, the existing gap between those who can experiment confidently and independently and those who cannot is widened.
Without constructive guidance from universities, students with external or informal support networks will be in a better position to test and use AI tools productively. Those without this existing support are more likely to struggle. Because they have no access to any guidance, they are more likely to misuse AI tools or rely on the output without fully understanding it. This results in fragmentation and increases existing inequalities.
Evaluation reform requires evaluating not only products but also processes.
Evaluation is another area where the wrong approach to AI is a challenge. Currently, assessment models are based on the outdated assumption that examining the work product (essay, coursework, etc.) will reveal whether a student understands the subject matter. With the advent of generative tools, that assumption breaks down.
Rather than retrofitting old structures, higher education’s experience with practical, competency-based applied assessment offers a potential new approach adapted to the age of AI: process, reasoning, and accountability assessment.
This can take several different forms. First, evaluate your decision to use the tool. In other words, deciding when to use AI and, if so, which tools are best suited for each use case. This may include research, analysis, copywriting, and editing. Another area is workflow documentation and evaluating how students are tracking their methods. It may also include a critical evaluation of AI-generated materials and consideration of where human judgment has intervened.
This shift in focus (from detecting misuse to fostering responsible practice) will create explainable AI-literate learners. why and how They used the tool in addition to submitting their completed answers.
A cross-cutting approach is more powerful than a piecemeal response
The broad lessons for this field are clear. The conversation needs to shift from “How do we stop AI?” “How do I teach my students to use it successfully?” This shift won’t be an option for long. As AI becomes embedded in various industries, employability, productivity, and career advancement will increasingly be determined by those who can use AI effectively and those who cannot.
Students pursuing further education have the opportunity to take a leadership position in this regard. By setting clear expectations, incorporating judgment-driven features, and aligning assessments with real-world practices, you can establish a model that the rest of the education ecosystem can follow. Otherwise, we risk a generation of learners who earn qualifications but fail to demonstrate what they can actually do.
The question is no longer whether AI belongs in education, but whether education systems and authorities can keep pace with the way AI is already reshaping learning and work.
From Dr. Paul Jung, CEO and Co-Founder of Medly
