Newsletter December 2025: TAL2025 – Teaching with AI
In this month’s newsletter, we once again turn our attention to GenAI, this time with inspiration from the recently held TAL2025, focusing on on how generative AI (GenAI) can be used critically and constructively in teaching.
On 6 November 2025, SDU CTL once again hosted the annual Teaching for Active Learning (TAL) conference for staff from SDU, University Colleges in the Region of Southern Denmark, and colleagues from other national universities and educational institutions. The conference featured 20 short communications, workshops, and posters, as well as two keynote presentations by José Antonio Bowen. Below are some selected insights from the event.
In addition to the main theme of active teaching and active learning, this year’s conference explored the topic of teaching with AI, addressing questions such as: How can we use AI critically and constructively to support learning, feedback, and supervision? How can pedagogical thinking guide AI use? And how can teaching with AI be supported through organisational structures and competence development?
In his two insightful and tightly packed keynotes – ”Teaching and Thinking with AI” and “Educating Humans to Thrive in an AI World” – José Bowen covered a wide range of perspectives. Below are a few selected points.
1. Chatbots, APIs, and prompts
Although GenAI has been publicly available for only three years, it no longer makes much sense to speak of “GenAI” in the singular. There are now several major competing language models on the market, and even more so-called APIs (application programming interfaces) designed for specialised use.
AI competence now also entails being able to select tools appropriately for different purposes. And there is growing collaboration around sharing effective prompts and prompting techniques, including for teaching purposes.
José Bowen’s resource page (https://weteachwithai.com/) provides valuable guidance for those wishing to navigate this rapidly evolving field. Below are a few selected points.
2. GenAI as a new condition in higher education – AI literacy as a new educational mission
Another key point from Bowen was that regardless of our position on GenAI, it has quickly become an unavoidable condition of modern teaching. It is now the responsibility of teahcers to help students use GenAI critically, with academic integrity and as a tool for learning.
AI literacy involves:
- the ability to ask good questions, and
- the ability to recognise limitations and critically evaluate AI outputs.
Bowen described prompting as a new form of academic writing: the more precise, nuanced, and well-structured a prompt or a dialogue sequence is, the greater the chance of receiving a meaningful response. “Prompting is writing” – and, in the university context, prompting is also thinking.
He also underscored the distinction between novices and experts:
- Novices often use AI to ‘figure out what to do’ but lack the ability to assess the quality of responses;
- Experts, on the other hand, use AI to accelerate what they already know and can critically revise, challenge, and refine its outputs.
Our educational task, then, is to cultivate experts – learners equipped to use AI responsibly and insightfully. AI literacy thus becomes a new form of educational formation, closely aligned with traditional goals of critical thinking, source criticism, and academic integrity – not something apart from disciplinary learning.
At the conference, the Danish School of Media and Journalism presented their work on strengthening students’ AI literacy and fostering reflection in teaching. Their survey of 200 students shows that AI is rarely discussed in terms of how and why it is used in group work.
Read a summery of their presentation on the role of AI literacy in collaborative work.
3. AI separates writing from thinking – teaching must reconnect them
A recurring theme in Bowen’s presentation was that AI increasingly separates writing from thinking: text of high linguistic quality can now be generated within seconds, without the writer necessarily undergoing any meaningful cognitive process.
If teaching and assessment continue to reward only the finished product (report, essay, or assignment), we risk assessing AI’s linguistic proficiency rather than the student’s understanding.
Bowen emphasised the need to design assignments where the process takes centre stage. This might involve:
- dividing writing tasks into distinct phases (idea development, outline, first draft, revision, reflection) and requiring documentation of where and how AI has been used;
- allowing students to use AI for restructuring or language editing while insisting that analytical choices, argumentation, source work, and disciplinary reasoning remain their own;
- encouraging meta-reflection: Did AI help you understand the material better – or merely make your writing more polished?
As Bowen put it, the goal is not to ban the tool, but to ‘reconnect thinking, writing, and AI use’ in ways that make clear that what is being assessed is the thinking itself.
In his book Teaching with AI, 2nd Edition, Bowen provides numerous examples of how such assignments can be designed.
At the conference, Lukas Galke, associate professor at SDU's Department of Mathematics and Computer Science, presented his experiences from a teaching activity where students work with LLM-assisted reading of research articles.
Read more about the LMM-Paperstorm teaching activity.
4. From “cheating” to academic progression – transparency in AI use
According to Bowen, the difference between “cheating” and “progression” often comes down to framing. Many activities that can now be automated (such as drafting, translation, or proofreading) were once considered signs of diligence – but in an AI context, these tasks can no longer constitute learning outcomes in themselves.
Teachers should therefore clarify which parts of an assignment may (or must) be completed with AI, and which represent the authentic academic performance.
Bowen suggested that we move from hidden to transparent AI use, where students document how they used AI and how it influenced their work.
This could include short reflections, documentation of prompts, or comparisons between AI-generated suggestions and the student’s own solutions.
In this way, AI becomes not a shortcut that bypasses learning, but a prompt for explicit discussions of disciplinary quality, ethics, and source critique.
5. AI as the new “average” – everyone becomes an AI manager
Bowen described how generative AI is rapidly becoming the new “average”: a baseline level of problem-solving, text production, and analysis that is now accessible to both students and teachers.
This means that many standard tasks – such as summaries, first drafts, and simple analyses – can be produced faster and more cheaply by AI than by humans.
Rather than competing with this new “average worker,” our challenge is to determine which aspects of academic work can be delegated to AI and where human judgement, ethics, creativity, and relationships remain essential.
A central insight was that AI constitutes a new form of labour: everyone – including students – has in effect become an “AI manager,” responsible for delegating tasks, evaluating outputs, and taking ownership of final products.
This increases the importance of viewing teaching not merely as a matter of content and methods, but also as workflow design: Which steps in an assignment can be supported by AI, and where must the student think, decide, evaluate, and take ownership?
6. AI as a driver of active learning – from lecture to simulation and dialogue
Bowen’s keynotes were rich in hands-on suggestions for how AI can support active learning before, during, and after teaching – in line with the core principles of the TAL conference.
Rather than treating AI solely as an information source, he proposed using it as an interactive learning partner, for example through:
- Simulation and role-play bots, where students engage with a “counterpart” (patient, client, policymaker, historical figure) and later analyse the interaction professionally.
- Bot assistants, tightly instructed by the teacher to pose specific types of questions, challenge assumptions, or provide discipline-tailored counterarguments.
- Reflection and feedback bots**, which help students articulate their learning, identify misconceptions, and generate new questions to bring back into class discussions.
A central slogan was that “every assignment is now an AI assignment” – not because AI must always be used, but because teachers should always consider what role AI plays (AI-resistant, AI-transparent, or AI-inclusive).
For the TAL conference’s focus on active learning, this means that AI is not just a technological topic but a concrete didactic strategy for promoting dialogue, multiple perspectives, and stronger connections between theory and practice.
At SDU, these ideas align closely with the university’s core principles of active learning and with the ongoing work to develop institutional guidelines for the use of GenAI in teaching and assessment.
TAL2025 demonstrated that many teachers are already experimenting in this space – and Bowen’s keynotes provided a shared vocabulary and pedagogical compass for engaging with AI in both critical and constructive ways.
The TAL conference also featured practical examples with a broader focus than GenAI. Flemming Smedegaard and Maria Mejnborg Lidsmoes presented findings from an ongoing project aimed at uncovering why collaboration among students often fails—and how universities can improve it.
