FE has a Chance to lead on Responsible AI Capability

Students are not alone in having quickly embraced generative AI. But in academia, it proposes a potential conundrum: should educational authorities’ response centre on how to restrict students’ AI use, or on educating students on how to use it responsibly?

Across the UK, responses to this question have been fragmented. Early findings from the Department for Education’s AI early adopters programme show how schools and colleges are already experimenting with different approaches, ranging from cautious restriction to active encouragement. This is normal with any widely adopted new technology – after all, many educational institutions are still grappling with mobile phones. 

But the consequence of this patchwork is that disparate adoption patterns among students have emerged: some students are learning to use AI tools productively and responsibly; some risk outsourcing their thinking to them altogether; and some aren’t using it all. This matters because it leads to inequity of education.

In this context, the further education sector has an important potential role to play in defining a better approach. It sits at the intersection of academic learning, technical skills and direct employer input, positioning it well to lead development a wider definition of constructive AI use. 

That means moving beyond both unhelpful outright bans and vague ideas of “responsible use” towards clear standards and forms of assessment that make students’ work credible and comparable. If that can be established here, it sets a precedent not just for other parts of education, but for how institutions and employers recognise and trust AI-augmented skills.

AI capability is becoming a core employability standard

Aside from improving students’ learning, getting AI use right matters because employers across many sectors are increasingly expecting recruits to work confidently with AI. AI fluency is becoming comparable to digital literacy a decade ago: a baseline expectation rather than a specialist skill. In practice this looks like employers expecting certain tasks to be delegated to AI, irritation with slow, repeated, manual workflows, and corresponding expectations of higher productivity. 

The necessary AI literacy for this changed landscape comes in two layers: knowing how to prompt and get the most out of AI as a tool, but also exercising judgement about when to use it (and not). The first of these is fast and easy to learn, but judgement, as in any field, only comes with extended learned experience.

This creates a tension for further education providers. They must maintain confidence in how students are assessed, while also preparing them for a workplace in which AI use is routine. Without the ability to interrogate outputs, explain reasoning, and validate results, learners risk entering the workforce with skills that don’t match expectations.

A better approach would be to move the focus away from whether students should use AI, and towards how they use it in ways that support learning. That means setting expectations, teaching judgement, and aligning assessment with real-world practice. This could strengthen both learning outcomes and the credibility of qualifications.

Prohibition creates inequality, structured guidance creates fairness

Another unavoidable problem with blanket bans as the default approach to AI is that they do not eliminate use. Instead, they push it into unstructured and unsupervised contexts, where learners are left to navigate varied and powerful tools without guidance. The consequence is that existing gaps between those who are confident experimenting independently and those who are not widen.

Without constructive guidance from colleges, students with outside or informal support networks will be in a stronger position to test and use AI tools productively. Those without such existing support are more likely to struggle – they won’t have any access to any guidance at all, meaning they’re more likely to misuse AI tools, disengage, or rely on outputs without fully understanding them. The result is fragmentation and an amplification of existing inequity.

Assessment reform must reward process, not just product

Assessment is another area where ill-conceived approaches to AI pose a challenge. Today, assessment models rest on an outdated assumption: that examining outputs (e.g. essays, coursework) reveals whether students understand subject matter. With the advent of generative tools, that assumption breaks down. 

Rather than retrofitting old structures, further education’s experience with practical, competency-based and applied assessment offers a potential new approach, adapted to the age of AI: evaluating process, reasoning and accountability.

This could take a few different forms. First, assessing decision-making about tool use; in other words, judgement on when to use AI, and if so, which tool is best suited to each use case. This could include research, analysis, copy writing and editing. Another area would be documentation of workflow, assessing how students are tracking their method; it could also include critical evaluation of AI-generated material and reflection on where human judgement intervened.

This shift in emphasis – from detecting misuse to nurturing responsible practice – would result in AI-literate learners who can explain why and how they used a tool, instead of just submitting a finished answer.

A cross-sector approach will be stronger than fragmented responses

The wider lesson for the sector is clear: the conversation should shift from “How do we stop AI?” to “How do we teach students to use it well?”. That shift will not be optional for long. As AI becomes embedded across industries, the gap between those who can use it effectively and those who cannot will increasingly determine employability, productivity and career progression.

Further education has an opportunity to take a leadership position on this. By setting clear expectations, embedding judgement-led capability, and aligning assessment with real-world practice, it can establish a model that other parts of the education ecosystem can follow. If it does not, the risk is a generation of learners whose qualifications fail to signal 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 how it is already reshaping learning and work.

By Dr. Paul Jung, CEO and Co-founder at Medly