Welcome back to the AI Commons. This fortnightly newsletter is funded by the Institute for Teaching and Learning and run by Mark Carrigan and Eva Parr. We're exploring how colleagues are integrating AI into teaching and research, with the aim of supporting learning across the university so we can approach this change in thoughtful, creative and collegial ways. We've had a number of colleagues get in touch recently to share what they've been doing, so this issue brings together several case studies from across the university.

📝 First, Dave Hirst (Library) describes an activity where students use Copilot to draft their coursework and then critically analyse what the tool actually produced

Digital Society is a University College of Interdisciplinary Learning unit developed and run by the Libraries Teaching and Learning and Students team.

In the Critical Analysis in a Digital World week**,** students focus on the importance of critically questioning the digital technologies that shape their lives, learning, and worldviews. The structure emphasises students' reflections on how they critically approach digital platforms, online information, AI-generated content, and the wider sociotechnical systems that influence behaviour and thinking online.

The content outlines what critical analysis means in a digital context: recognising assumptions, identifying bias, evaluating sources, understanding how platforms shape what users see, and questioning how algorithms influence decisions. It emphasises that digital media is not neutral: it steers attention, prioritises certain narratives, and determines what is visible or invisible. Students are asked to reflect on how they engage in their digital environments.

As the unit lead, I have noticed the negative impact of GenAI tools on student assessment and wanted to build in an activity to encourage the cohort to evaluate and reflect on using these tools in their coursework.

The activity asks students to use a GenAI tool to help write their first assessment and then critically analyse that process. We ask them to use Copilot as the University's recommended tool. The activity is scaffolded to ensure deep reflection rather than passive or uncritical use of GenAI. Students are asked to generate an initial AI-written response in Copilot and then refine their prompts to see how different instructions influence the tool's output. Results are compared, emphasising the importance of noting changes in accuracy, tone, structure, argumentation, and use of evidence. The next step is critical reflection: how does GenAI construct knowledge? How do its limitations affect learning and academic integrity? Finally, they explain their own use of GenAI, outlining what they accepted, rejected, modified, or rewrote.

The activity is not simply "use AI" but takes a structured approach ensuring students reflect on the "thinking" the AI tool appears to do. Students are asked to adjust prompts, compare outputs, critique the model's behaviour, and justify their decisions. They are positioned as evaluators, not users outsourcing their work. By requiring explicit reflection, the activity ensures students recognise how GenAI tools construct text and how prompts influence content and quality. It also asks them to consider what GenAI gets wrong and why, and how their own judgement can shape their academic work.

My hope is that the activity transforms AI from a shortcut into a learning tool; one that reveals the mechanics and risks of these systems while developing critical digital literacy. One student put it well: "helpful in terms of structure and organising key points but there were clear limitations. Outputs lacked nuance, criticality, references could not be verified and under close scrutiny, surface level analysis fell apart. Quality academic work requires human judgement and verification through credible academic sources."

📝 Second, Nahielly Palacios (Education) reflects on a tutorial sequence where students designed their own language learning materials then compared them with what Gemini produced:

As the course director of the unit Digital Literacy in Language Teaching I have embraced the challenges posed by Generative AI in Language Education. The unit now incorporates a series of lectures and tutorials where students explore the use of different models, learn to write prompts, and develop their critical AI literacy.

This year, I ran a couple of tutorials where I asked students to create their own language learning materials and then compare them with materials generated by Gemini. The purpose of this sequence was threefold: to give students experience designing materials, to explore how these tools can support this process and to put into practice AI literacy skills by evaluating AI-generated outputs.

In the first tutorial, I asked students to work collaboratively to design a language learning material from scratch. For this, students identified a teaching context where they could apply this material, developed language learning outcomes, looked at authentic materials as well as didactic materials for inspiration and designed a sequence of activities. Each group presented their design the following week, explaining their decision-making process, drawing on the literature that informed such decisions and articulating how their design reflected their pedagogical values and teaching philosophy. During this activity, I noticed students having lively discussions, asking each other questions and building on each other’s ideas as they developed and shared their designs.

In the second tutorial, I asked the same groups to use Gemini to complete a similar task. For this, students applied prompting strategies explored earlier in the course: role assignment, task description, tone, target audience, teaching context, and intended learning outcomes. Once an output was generated, I encouraged students to evaluate it, discuss their findings and decide whether they would use the output and why. For instance, students examined the appropriacy of the design in relation to learners’ needs, teaching context, and teaching approach. Some students noticed inconsistency in the proposed activities and their pedagogy. Other students saw potential in how quickly Gemini could create a structured lesson. Students’ discussions showed deep understanding of the challenges of relying on AI tools and the relevance of not accepting AI outputs at face value. Many preferred their own design, appreciating their pedagogical values, teacher identity, creativity and collaborative work.

Reflecting on this teaching experience, I can see how valuable it is to get students to compare human-created with AI-generated materials, especially if they are involved in the development of both, as this process can help them recognise their own voice as teachers and material designers while developing and enhancing their AI literacy skills. Going forward, I plan to continue using this approach encouraging students to engage in deeper reflection on how AI can support their teaching without replacing their creativity, teacher identity, and understanding of their teaching context.

📝 Third, Vlad Porumb and Anne Stafford (AMBS) describe how they've redesigned a 500-student accounting course to integrate generative AI into teaching, learning, and assessment:

Artificial intelligence is reshaping education at remarkable speed—and accounting education is no exception. Here we share how we embed generative AI as a proactive pedagogical innovation for teaching, learning, and assessment in a 2nd year UG Financial Statement Analysis course of around 500 students. 

Generative AI is disrupting accounting education, posing challenges and opportunities for educators. Three key contextual factors shape our response: online, open-book assessments enable authentic designs beyond professional accreditation constraints; rapid technological change in accounting firms requires students to graduate with proficient, ethical AI skills; and student behaviour varies, with some using AI extensively without full awareness of the risks and others avoiding it due to plagiarism concerns. 

Accounting educators have a responsibility to prepare students for a professional landscape shaped by AI. This means adapting our teaching to equip graduates with the necessary skills for competent and ethical AI use, promoting awareness of its risks and advantages. 

We have shifted from traditional lecturer-led learning to generative AI-enabled approaches, purposefully integrating AI into the curriculum. We introduced an AI Code of Conduct, clarifying ethical use, plagiarism risks, and best practice. We redesigned assessment to deter copy‑paste AI misuse, replacing elements like ratio calculations and theory essays with tasks requiring rationale, stakeholder recommendations, and reflections on generative AI use. We also revised grading methods to match evolving generative AI tools. 

More radically, we now integrate generative AI into learning as a dynamic, adaptive process. Playful experimentation in class allows students to use generative AI for deeper financial analysis, such as examining the profitability of Shell, or considering the perspective of a villager in the Niger delta affected by oil spillage. “What if” questions are explored, like changes in profitability due to oil price fluctuations or Shell’s responses to climate change scenarios. This collaborative approach encourages curiosity and richer discussions between students and educators. 

A pivotal “aha moment” came when we recognised the value of designing our use of generative AI to be more proactive and curious, further nudging student engagement. Now we prioritise critical thinking and technical expertise before integrating generative AI tools, enabling ethical experimentation and prompt engineering. Students assess their own work through self-feedback and receive personalised feedback, which they report as clear and easy to follow. Anecdotal evidence suggests students who engage deeply and use critical thinking, targeted prompting and self-feedback techniques perform very well, while others who do not prepare or practise sufficiently lag behind.  

Our approach demonstrates how the thoughtful integration of generative AI tools can boost engagement and learning outcomes by fostering active interaction between learners, educators, and AI itself. This encourages discussions from multiple viewpoints and provides immediate, personalised feedback. Such engagement supports deeper learning and skill development while minimising the risk of disengagement sometimes associated with generative AI. In particular, using these dialogic methods, where learners and educators engage in ongoing, two-way conversations and collaborative exchanges, make assessments more robust, as they require students to articulate their reasoning and respond to questions, making it harder to rely solely on AI-generated answers. 

Although our ongoing journey is daunting at times, and requires rethinking around student preparation, teaching strategies and assessment design, overall we’ve found it to be a rewarding process.  

📝 Fourth, Bean Sharp (Library) on a Digital Society topic that asks students to consider who benefits from GenAI, who is harmed, and who gets to decide:

A recent article by Mancunian Matters on the University’s new Microsoft partnership shows that our students have thoughtful questions about generative AI (GenAI) tools and what it means to engage with them ethically. The University College of Interdisciplinary Learning (UCIL) unit Digital Society (also discussed in unit convenor David Hirst’s case study earlier), offers students space to reflect critically on these concerns, drawing on their own knowledge and experience with technology.

This week of the unit explores how GenAI tools might extend our capacities for ethical intervention while also producing new forms of harm. How do we distinguish between GenAI tools that might spot signs of cancer more accurately than a human and those that normalise copyright theft, perpetuate hate speech, exploit workers, and accelerate climate collapse? The AI, Ethics and Us week helps students develop nuanced and considered positions on GenAI usage and regulation.

We interrogate what it means to control the unethical outputs of some GenAI tools, given these tasks carry their own ethical quandaries. Who decides which content is unethical? Who will have to sift through violent, hateful, and disturbing content to ensure end-users aren’t exposed? And who should be held accountable for dangerous or damaging GenAI outputs? These are challenging real-world problems that need urgent attention.

Digital Society students offer multidisciplinary perspectives on these problems, as cohorts that are increasingly told they will need to embed GenAI tools in their learning, working, and daily lives. Among the primary target audiences of GenAI, students know first-hand the draw of these tools in terms of time saving and convenience. However, students are also deeply conscious of the expectations they’re burdened with, as the workforce of tomorrow, to tackle current and future crises including rising authoritarianism, widening inequality, and our ever-warming planet. Will GenAI aid us in tackling these interconnected global catastrophes? Or will these tools entrench our predicament or introduce new problems?

Through interactive comment boxes and polls, students share their responses and developing thought processes as they progress through the week’s material. We designed activities that would allow students to learn through experimenting with GenAI tools, guiding them to think critically about how these tools work and the hidden labour behind their outputs. For example, to get students thinking about the copyright issues relating to GenAI tools, we incorporated an activity that asks students to use an image generation tool to create something in the style of an artist of their choice. Students then reflect on who should get credit for the image. Moving beyond the University academic integrity policy approach to GenAI tools, we ask students to consider in what ways GenAI tools might restrict their thinking, given the guardrails of these tools don’t map neatly onto the academic freedom they’re entitled to in a university setting.

In conjunction with the other topics on this UCIL unit, AI, Ethics and Us prompts students to resist the marketing hype of big tech and dig deeper into what it means to navigate our digital world.  

📝 Finally, Martyn Edwards (MIE) on why he stopped trying to design AI-proof assessments and what he does instead:

Like many colleagues, my first reaction to the arrival of LLMs, and its potential impact on assessment, was of great concern and I found myself asking “If students can generate answers at the click of a button, what exactly are we assessing anymore?”  But the more I thought about it, the more I realised this was not a new problem, just one now made harder to ignore - “If an assessment can be easily outsourced to AI, was it ever really capturing meaningful learning in the first place?”  So instead of trying to design AI-proof assessments (which I am not convinced is fully possible), I have shifted my approach towards more productive assessment designs that students are more likely to enjoy whilst assessing higher-order thinking skills that are difficult to outsource to an LLM. 

A lot of this comes down to how I choose to position assessment within my unit design. Traditionally, assessment can feel like something that happens at the end and is seen as a way of checking if learning has taken place.  However, I treat assessment as part of the learning process itself, positioning closely with what Biggs describes as constructive alignment, where teaching activities and assessment are deliberately designed to enable students to demonstrate the outcomes we want them to achieve. In practical terms, this means moving the focus away from “what is the final answer?” to something more meaningful: “how and why did you get your answer?” 

In one module students are introduced to Bourdieu’s ideas around cultural, social, and symbolic capital. Rather than asking the students to undertake a traditional essay, they were challenged to apply these conceptual ideas to a real-world scenario: what would happen if a street in Manchester was turned into a museum? This assessment required the students to consider different perspectives regarding their choices, thinking through potential consequences and offering potential mitigations, and ultimately come to a reasoned judgement based on their research. Tasks like this draw on what Wiggins would describe as authentic assessment which allows students space to perform complex, meaningful tasks rather than simply recall knowledge out of context. 

This approach naturally pushes students towards higher-order thinking, resulting in students not just describing theory, but applying theory and considering its impact on real-world problems.  In Bloom’s terms, they are moving beyond remembering and understanding into analysing, evaluating, and creating and these types of thinking processes that are much harder to outsource, because they place judgement, context, and decision-making at the centre of the assessment strategy.  

This links closely to motivation, and by extension, the acceptance that students are motivated in very different ways.  For some it’s curiosity, others it’s fear of failure, and for others by the desire to progress, but what consistently makes a difference is whether they see value in what they are doing. Self-determination theory highlights that intrinsic motivation, doing something because it is meaningful or interesting, is far more powerful than working for grades. If an assessment feels like something worth doing, students are more likely to engage with it differently.  

Feedback is also central to my approach, and I put a lot of emphasis on formative feedback throughout the process. Conversations, questions, small interventions. This shifts the emphasis away from producing a final product and towards developing ideas over time. We have known for some time that this kind of ongoing feedback can significantly improve learning outcomes, particularly when students are actively involved in reflecting on and shaping their own work.  

So where does AI fit into all of this? For me, it becomes less of a threat and more of a background factor. If students are being asked to apply ideas, make decisions, reflect, and engage in dialogue, then AI has far less space to do the heavy lifting.  And so, I think it comes down to trust, in both themselves, and in their lecturers/markers. If we want students to engage meaningfully, we must design assessments that are worth engaging with. That might mean sacrificing some efficiency (more conversations, more iterative work) but the payoff is a much clearer sense of what students understand, and why and how they can use it in the future.  The focus is still the same as it should always have been: helping students think, apply, question, and create in ways that feel meaningful to them. In doing this AI becomes just another tool, albeit a powerful one, to help humans think, design, collaborate, and produce artifacts of value and worth. 

☕️ Workshop: Developing Student Awareness of AI

Wednesday 15 July, 1–3pm | Chemistry G.54 and online via Teams

Most students will already be using generative AI when they arrive at Manchester, but many won't have a good understanding of how to use it ethically and proficiently. This workshop explores how we can embed guidance—from plagiarism advice to understanding AI's limitations—into our programmes, and how to develop student critical thinking skills in response to these challenges.

Moderators: Daniel Engstrom and David Schultz
Panellists: Cesare Giulio Ardito, Katie Moore

We hope you enjoyed this sixth issue of the AI Commons. If you found it valuable, would you consider forwarding this newsletter to your colleagues? Comments, suggestions and questions are always welcome.