This month in the AI in FE community, we were joined by Penny Langford, Executive Head of Digital at Milton Keynes College, for a hands-on session focused on AI literacy. Penny took us through the kinds of practical activities they use with students and also shared with us some of the work they have been doing to develop and test AI agents.
Continue for a recap of the session and discussions, you can also access the full recording and transcript on our YouTube channel.
How many students are using AI?
Penny opened the session with some questions for us, beginning with: What percentage of your students are using AI? Responses leaned heavily towards “most students”, with the majority selecting the 50 to 99 percent range and a few selecting 100%.
This prompted an interesting discussion about what using AI actually means right now. Members pointed out that students may not be intentionally opening a chatbot, but AI features are already embedded in search, apps, and recommendations on their phones. Penny shared that when she asked students this question directly, the numbers using AI were low, there’s a feeling that students still are not comfortable admitting AI use.
How Milton Keynes College is embedding AI literacy
Penny then shared how AI literacy is being made visible and measurable across the college.
AI literacy is being built into schemes of work, and into a quality scorecard process aligned to the new Ofsted framework, wherein curriculum areas are rated against a set of benchmarks each half term in which they’ll include AI literacy.
She also talked through how the college has rolled out Copilot for students aged 13+ this year. This has been done carefully, including consent and ethics considerations, review through an ethics board, and thorough testing. Staff and student AI literacy training has been delivered alongside the rollout.
Another key part of their approach is making guidance easy to access. Penny shared their internal “Everything AI” space for staff, and a parallel area for students. This includes an AI literacy framework, guidance around assessment, tool guidance, and activity ideas mapped to the framework. Penny noted they have adopted the Jisc AI literacy framework as the basis for this work.
Practical activities for AI literacy
A big part of the session was trying out activities in the same way Milton Keynes College uses them with staff and students.
We started with a quick “AI or not AI” sorting activity. Penny shared that they use this with learners as a way of exploring what counts as AI, what does not, and what might surprise students. The activity also works well as a research task, where learners validate answers and then create their own versions, for example using Canva code or Microsoft Forms.
We then moved into a short discussion of how students are already using AI, with responses including academic activities like planning, research, and summarising, but also wellbeing and emotional support. Penny shared that when their team tested Copilot for 13+, they specifically explored wellbeing-related prompts and escalation routes. One useful insight was that their own safeguarding monitoring processes worked quickly during testing, and the guardrails in Copilot itself were encouraging.
Exploring approaches to prompting
Penny also shared a light-touch but effective approach they use with staff and students to build understanding of prompting, including a fun exercise generating brick figures as a way to practise describing detail and being specific.
This led us into our second activity where we were tasked with comparing outputs from the same prompt, evaluating the differences, and building a group response. Members noted how the activity gave hands on experience of how varied outputs can be even when the starting prompt is similar, and how there is a balance to strike detail to provide in the prompt – some were surprised to see how tools would guide the user forward even when prompt was incomplete.
Penny emphasised that AI literacy is not just about showing learners how to get a good answer from an AI tool. It is also about helping them take the next step to think critically about the output too, asking questions like: how was this produced, what is it based on, what is missing, and how could it be improved?
Addressing ethical scenarios
In our final activity, Penny introduced us to a series of ethical scenario cards each touching on key issues in AI ethics from facial recognition and privacy to AI use in recruitment and bias. Each scenario came with questions designed to spark discussion about responsible use, risks, and where boundaries sit.
Members explored the scenarios in groups and also experimented with generating their own subject-specific scenarios using AI tools, thinking about how to bring these conversations into the context of different subjects rather than treating AI ethics as a standalone topic.
A look at AI agents
In the final section, Penny shared some examples of Copilot AI agents the college has been developing.
One focus has been creating an agent to support teaching staff with lesson ideas based on the college’s teaching framework, alongside inclusion and SEND-focused material. They have also been building agents for internal support, with teams creating agents based on supporting documents from their intranet.
Penny also described a more sophisticated course enquiry agent for prospective students built in Copilot Studio, and a GCSE revision agent developed from an existing quiz-maker template which is now being tested by English and maths staff.
Testing, governance, and cheesecake
A key takeaway from this part of the session was Penny’s emphasis on testing agents.
Milton Keynes College has developed a testing framework, aligned with wider cyber and information security expectations. Penny highlighted several practical safeguards, including restricting agents to approved sources rather than allowing open internet access, and thinking carefully about where automations and agents “live” so they are not tied to one individual’s account if that person changes roles or leaves.
Penny ended the session sharing a simple test that acts as a quick sense-check – asking the agent “how do you make cheesecake?”. If an agent has been locked down to its intended knowledge base, it should not be able to answer.
Thank you again to Penny for a session that was genuinely interactive and full of brilliant, practical ideas.
Join us in March
Our next AI in FE community session will be on 17 March – 1pm to 2pm. We will be joined by Craig Scott, Welding / Hydrogen Tutor from Riverside College. Craig will be sharing a practical FE case study from a class preparing for an online exam using different AI methods.
As always, sessions are open to Jisc member institutions, join our jiscmail list to receive invites.
If you would like to present a future session, whether that is a case study, discussion topic, or something you are piloting, please get in touch via ai@jisc.ac.uk.
Find out more by visiting our Artificial Intelligence page to explore publications and resources, learn more about our communities and sign up for our AI Literacy training.
For regular updates from the team sign up to our mailing list.
Get in touch with the team directly at AI@jisc.ac.uk