When generative artificial intelligence leapt from research labs into everyday life,universities worldwide were forced into a rapid reckoning: was this a threat to academic integrity or a powerful new instrument for learning? At the London School of Economics and Political Science (LSE),that question has given rise to GENIAL – “GENerative AI Tools as a Catalyst for Learning” – an institution-wide initiative that aims to move beyond alarmist headlines and harness AI as a constructive force in higher education.
Positioned at the intersection of technology, pedagogy and policy, GENIAL is LSE’s attempt to systematically explore how tools such as ChatGPT, Claude and other large language models can reshape teaching, assessment and student support. Rather than banning or blindly embracing these systems, the project is testing their limits in real classrooms, interrogating their implications for critical thinking and academic rigour, and asking what “learning” means when machines can generate passable essays in seconds.
Drawing on LSE’s strengths in social science, ethics and public policy, GENIAL is not just a technical experiment but a live case study in how universities can adapt to – and help govern – the AI age.
Unlocking the potential of generative AI at LSE as a catalyst for deeper student learning
On a campus where critical inquiry is a defining tradition, generative AI is emerging not as a shortcut, but as an academic sparring partner. At LSE, the GENIAL initiative is reframing these tools as companions for reflection, iteration and critique. Rather than producing finished answers, AI becomes a sandbox in which students can test hypotheses, expose hidden assumptions and rehearse complex arguments before committing them to paper. In seminars and independent study alike, students are experimenting with AI to simulate policy negotiations, unpack econometric results and rehearse interview-style questioning that deepens their grasp of theory and evidence.
This shift is being carefully structured so that experimentation enhances, rather than erodes, academic integrity and intellectual rigour. Teaching teams are co-designing activities where AI is explicitly positioned as a prompt for higher-order thinking, including:
- Pre-seminar briefings that use AI to generate contrasting summaries of core readings for students to critique.
- Policy scenario simulations where AI plays the role of stakeholders, encouraging students to probe trade‑offs and unintended consequences.
- Methodological “what if” explorations that let students quickly compare option research designs before conducting their own analysis.
| Teaching Aim | AI-Supported Activity | Student Outcome |
|---|---|---|
| Strengthen argumentation | Generate counter-arguments to critique | Sharper, evidence-based claims |
| Demystify complex data | Explain models in plain language | Greater conceptual clarity |
| Connect theory to practice | Create realistic case vignettes | Richer submission of frameworks |
Designing responsible AI use in the classroom from assessment policies to academic integrity
At LSE, the shift from banning to embedding generative AI is reshaping how assessment is designed, explained and defended.Course conveners are moving beyond vague “don’t cheat” statements towards assessment briefs that explicitly define acceptable AI support at each stage of a task, from brainstorming to proofreading. This includes clear references in marking criteria to how well students disclose and critically evaluate AI use,rather than simply penalising it. To support staff and students alike, departments are starting to publish transparent AI usage matrices that differentiate between formative and summative work, and highlight when AI is a learning partner versus when it risks eroding the evidence of a student’s own understanding.
- Clarify boundaries: specify which tools are allowed and for what purposes.
- Design AI-aware tasks: favour personalised, data-driven and reflective assignments.
- Reward transparency: assess students on how they document and critique AI assistance.
- Educate on ethics: link AI practices to academic integrity, bias and data privacy.
| Scenario | AI Use | Integrity Focus |
|---|---|---|
| Essay planning | Idea generation with citation checking | Source verification and correct attribution |
| Data analysis | Code suggestions in a sandbox | Understanding logic, not copying output |
| Group projects | Shared AI-assisted drafts | Tracing authorship and documenting roles |
| Take-home exams | AI use restricted or prohibited | Upholding individual, unsupplemented work |
Building digital confidence among staff and students through targeted GENIAL training and support
LSE’s GENIAL initiative recognises that digital confidence is not a static skill set but a mindset that can be nurtured through intentional practice, critical reflection and community support. Workshops are structured around live demonstrations, hands-on labs and reflective debriefs, enabling participants to move from passive curiosity to active experimentation with generative AI in their own disciplinary contexts. Tailored pathways for academics, professional services staff and students ensure that everyone encounters tools, prompts and ethical scenarios that resonate with their day‑to‑day realities. Short, focused micro‑credentials and drop‑in “AI clinics” reduce barriers to engagement, allowing individuals to test ideas, share concerns and receive immediate feedback from specialists and peers.
To make this shift enduring, GENIAL embeds support into existing learning and teaching ecosystems rather than treating AI as an add‑on. Staff development sessions are co-designed with departments so that teaching teams can collectively agree on expectations,assessment designs and governance. Students, in turn, are given structured opportunities to practise responsible use of AI through guided activities such as critical prompt engineering, bias detection and output verification. Across all formats, emphasis is placed on:
- Transparency – clarifying when and how AI is used in teaching and assessment
- Critical literacy – interrogating sources, limitations and potential harms
- Collaboration – sharing exemplars, failures and workarounds across the LSE community
- Accountability – aligning AI use with institutional policies and academic integrity
| GENIAL offer | Audience | Confidence gain |
|---|---|---|
| Scenario-based workshops | Teaching staff | Designing AI-aware assessments |
| Prompt labs | Students | Stronger critical questioning skills |
| AI clinics | All staff | Practical troubleshooting and peer advice |
From pilots to practice embedding GENIAL tools in curricula and measuring their impact on learning outcomes
The first wave of GENIAL experimentation at LSE began with tightly scoped pilots in seminar rooms, computer labs and virtual learning spaces, where academics could safely trial AI-supported activities alongside conventional teaching. These early tests helped clarify where generative tools enhanced learning – and where they distracted from it. As confidence grew, departments started weaving GENIAL into core course designs, aligning AI-driven tasks with existing learning outcomes rather than treating them as add‑ons. Lecturers collaborated with learning technologists to redesign assessments,feedback mechanisms and in‑class exercises so that AI tools supported critical thinking,methodological rigour and discipline‑specific skills.
To understand whether these changes genuinely improved student learning, LSE adopted a mixed‑methods evaluation framework that combined quantitative indicators with rich qualitative evidence. Course teams now regularly monitor:
- Assessment performance before and after AI-enhanced activities
- Engagement metrics in the virtual learning environment
- Student reflections on AI-supported tasks and feedback
- Academic integrity signals, including shifts in misconduct patterns
| Measure | What it shows | Early trend |
|---|---|---|
| Concept quizzes | Depth of understanding | Higher median scores |
| Draft submissions | Use of AI in revision | More iterative work |
| Student surveys | Perceived value of tools | Rising satisfaction |
| Office hours | Nature of questions | Shift to higher-order issues |
Key Takeaways
As generative AI moves from the margins of experimentation to the mainstream of higher education, LSE’s GENIAL initiative offers a glimpse of what a more deliberate, critically informed adoption might look like.Rather than treating these tools as a shortcut or a threat,the project frames them as objects of inquiry and instruments for learning – to be scrutinised,tested and debated as much as they are deployed.
The coming years will determine whether AI in universities merely automates existing practices or genuinely reshapes how students think, research and write. At LSE, the bet is on the latter: that with the right scaffolding, governance and pedagogy, GENerative AI can become a catalyst rather than a crutch. If that wager pays off,the lessons from this experiment in Houghton Street could travel far beyond the confines of one institution – helping to define not only how we use AI to learn,but what it means to be educated in an age of machines that can also generate text,images and code.
