Artificial intelligence is rapidly reshaping the way we teach, learn, and assess-and universities are moving fast to keep pace. At King’s College London, this change took center stage at the recent “AI in Education Showcase: Lightning Presentations,” where academics, technologists, and students gathered to share how emerging tools are already changing classroom practice.
Over a series of brief, focused talks, presenters demonstrated real-world applications of AI across disciplines: from automated feedback systems and adaptive learning platforms to AI-supported assessment and new forms of digital scholarship. Rather than dwelling on distant futures, the showcase highlighted what is happening now in lecture theatres, laboratories, and virtual learning environments at King’s.
As higher education confronts both the promise and the risks of generative AI, this event offered a snapshot of a sector in transition-experimenting in real time, debating ethical implications, and reimagining what a university education might look like in an AI-enabled world.
Exploring how AI tools are reshaping teaching and assessment at King’s College London
From lecture theatres in Strand to clinical labs at Denmark Hill, educators are quietly weaving generative tools into day‑to‑day practice, moving beyond hype into pragmatic experimentation. At this showcase, academics demonstrated how they are using AI to generate rapid formative feedback, simulate complex case studies and scaffold academic writing without diluting rigour. A law lecturer, as a notable example, uses a bespoke chatbot to present students with evolving legal scenarios, while a language tutor deploys AI‑driven dialog partners that adapt to each learner’s fluency level.These interventions are not replacing customary scholarship; they are restructuring how time is spent, allowing teaching staff to focus on higher‑order discussion and pastoral support rather than repetitive administrative tasks.
Assessment design is undergoing an equally significant recalibration. Instead of simply trying to “AI‑proof” exams, staff are experimenting with assignments that require students to critique, verify and improve AI outputs, making academic integrity an explicit learning outcome.Early pilots indicate that when students are guided to use AI transparently, their engagement and meta‑cognitive awareness increase. Key themes emerging from the lightning presentations included:
- Authentic assessment through real‑world tasks that integrate AI use
- Transparency in communicating acceptable AI practices to students
- Staff development that equips educators to critically appraise new tools
- Equity and access so AI does not widen existing learning gaps
| Area | AI Use at King’s | Impact |
|---|---|---|
| Formative feedback | Draft‑review assistants | Faster, more targeted comments |
| Clinical training | Simulated patient cases | Safer practice, richer scenarios |
| Essay assessment | AI critique tasks | Stronger critical thinking |
| STEM learning | Code clarification tools | Deeper conceptual grasp |
Inside the lightning presentations what worked what failed and what surprised educators
In ten-minute bursts, lecturers pulled back the curtain on their experiments with AI tools, revealing a mix of quiet breakthroughs and messy misfires.Some described how AI-generated question banks let them diversify assessments overnight, freeing time for richer feedback and one-to-one supervision. Others showcased chatbot “study buddies” embedded in virtual learning environments, reporting that anxious first-years were far more likely to pose “silly questions” to an algorithm than to a tutor. Patterns quickly emerged:
- Worked: AI as a drafting partner for feedback comments and rubric design
- Failed: Over-automated marking that flattened nuance and tone
- Surprised: Students using AI to challenge model answers, not just copy them
- Worked: Multilingual support that narrowed participation gaps in seminars
- Failed: Tools introduced without clear guardrails or ethical framing
| Session Type | Big Win | Unexpected Insight |
|---|---|---|
| Assessment Design | Faster, more varied question sets | Higher-order tasks reduced AI overreliance |
| Student Support | 24/7 AI “office hours” | Increased confidence among quieter students |
| Academic Integrity | Transparent AI policies | Students requested co-created AI norms |
What resonated most was how quickly initial scepticism softened once educators saw AI augment, rather than replace, their professional judgement. Many reported that co-designing prompts with students turned potential conflict into collaboration, while a handful admitted that their first pilots failed simply because they had tried to “bolt AI on” instead of aligning it with existing pedagogy. Across subjects and formats, the presentations underscored a shared lesson: the real innovation lay less in the tools themselves, and more in how human expertise, critical reflection, and classroom culture shaped their use.
Balancing innovation and integrity integrating AI in the classroom without compromising academic standards
At King’s, lecturers are beginning to treat AI tools as they would any powerful research database: valuable, but never a substitute for original thought. Rather of banning generative platforms, modules are redesigning tasks so that students must critique, adapt and refine AI-generated material rather than submit it wholesale. Seminar leaders are asking students to submit short process reflections alongside essays, explaining which prompts they tried, how they evaluated the output, and where they chose to depart from it. This makes academic integrity visible as a skill,not just a rule,and it helps staff distinguish between genuine scholarship and over-reliance on automation.
- Clear guidance on permitted AI use in each assessment
- Assessment briefs that foreground critical evaluation over reproduction
- Workshops on bias,hallucination and citation ethics
- Routine use of drafts,peer review and viva-style conversations
| AI Classroom Practice | Integrity Safeguard |
|---|---|
| AI-assisted brainstorming | Student logs ideas kept,changed,or rejected |
| Machine-generated summaries | Spot-checks against set readings in seminars |
| Draft feedback from chatbots | Lecturer assesses revision trail,not just final text |
Crucially,course teams are embedding transparent alignment between learning outcomes,AI policies and marking criteria. Rubrics now specify where self-reliant analysis, use of primary sources and accurate attribution carry the most weight, signalling that shortcuts via chatbots will simply not score well. In parallel, departments are developing shared FAQs and model statements students can paste into their work to disclose when and how they used AI. Rather than policing in the dark, King’s is normalising honest reporting, equipping students to navigate emerging technologies while preserving the rigour, authorship and trust on which academic communities depend.
From pilots to policy concrete recommendations for scaling responsible AI in education at King’s College London
Across departments, experimental classroom tools and niche research projects are now informing institution-wide decisions about how AI is designed, deployed and governed. Drawing on these pilots, King’s is outlining a practical playbook that moves beyond hype, centring equity, academic integrity and human agency. Emerging guidance focuses on three strands: clear role definitions for staff and students; transparent data practices; and disciplined evaluation of impact on learning,not just efficiency. These strands are beginning to shape assessment design, curriculum refresh cycles and new expectations for digital literacy that extend from first-year undergraduates to senior academics.
To turn these insights into action, King’s is crafting a set of concrete levers that can be embedded in institutional practice:
- Curriculum alignment: mapping AI use to specific learning outcomes and disciplinary norms.
- Staff capability-building: short, targeted workshops anchored in real teaching scenarios.
- Student co-creation: involving learners in testing and critiquing AI tools before adoption.
- Governance by design: lightweight review routes for new tools, with clear escalation paths.
- Transparency defaults: model cards, consent prompts and visible provenance of AI-generated content.
| Area | Pilot Insight | Policy Shift |
|---|---|---|
| Assessment | AI used as a drafting aid | Explicit rules on acceptable support |
| Feedback | Scalable formative comments | Human sign‑off for high‑stakes work |
| Access | Boost for neurodiverse learners | Inclusive defaults, not add‑ons |
| Data | Fragmented tool landscape | Approved, centrally supported platforms |
The Way Forward
As the session drew to a close, one message resonated across all the lightning presentations: AI in education is no longer a distant prospect but an active, evolving practice at King’s College London. From reimagining assessment and feedback to creating more inclusive learning environments, the showcased projects revealed both the ambition and the careful scrutiny with which staff and students are approaching these tools.
What emerged was not a narrative of technology replacing teaching, but of academics experimenting with new ways to enhance it-always framed by questions of ethics, equity and academic integrity.The showcase underlined that any prosperous adoption of AI will depend as much on dialogue, critical reflection and collaboration as on the technology itself.
As King’s continues to build on this work, the event served as a snapshot of a sector in transition: one where AI is being tested in real classrooms, refined through practice, and shaped by the people who will ultimately use it. The challenge now is to turn promising pilots into enduring, evidence-based approaches that keep human learning at their core.
