- Faculty/module/cohort (if applicable): Faculty of Arts, Science and Technology (FAST) – MSc Molecular Medicine (Dissertation Module) and Biological Sciences
- AI Tool(s) Used: ChatGPT (or equivalent generative AI), Blackboard, H5P, Python/R code generation
- AI mode(s) used (e.g., text → image, audio → text, multimodal prompts): Text generation, prompt engineering, AI-assisted instructional design, code generation, simulated data creation, feedback and critique.
What was the challenge?
Dr Lee Machado wanted to improve student engagement and supervision within the MSc Molecular Medicine dissertation module, where students complete an independent research project worth 70% of a 60-credit module. Although the dissertation spans several months, formal contact time with students is relatively limited, making it difficult to monitor the quality of supervisor interactions and identify students who may require additional support.
Alongside dissertation supervision, Lee also wanted students to develop stronger academic writing, critical thinking and ethical AI literacy while preparing them for professional practice in research and industry. Existing assessment approaches also required reconsideration in response to increasing AI use within higher education.
What did you do?
AI was used across several connected teaching activities.
AI-supported dissertation supervision
Using generative AI, Lee designed a structured supervisor interaction log within Blackboard using H5P. Through prompt engineering, AI suggested ways to build an interactive log that allowed students to record:
- Meetings with supervisors
- Laboratory activities
- Agreed actions
- Reflections on progress
- Outcomes from each supervision session
The log enabled students and supervisors to monitor progress while allowing the module leader to identify students who required additional support.
Supporting academic writing
AI was introduced during dissertation writing workshops to help students distinguish between different sections of scientific writing.
Students generated examples of both weak and stronger Results and Discussion chapters before critically evaluating the AI-generated responses. They then compared these examples against the expectations discussed during teaching sessions to identify strengths, weaknesses and opportunities for improvement.
Rather than producing work for students, AI acted as a critical discussion partner that encouraged deeper reflection on academic writing.
Assessment redesign
Lee also explored using AI-generated simulated biological datasets within assessments.
Instead of traditional essay titles, students were presented with realistic research data and asked to take on the role of medical writers preparing reports for pharmaceutical companies. This created a more authentic assessment that better reflected professional practice while encouraging students to interpret evidence rather than simply reproduce knowledge.
Multimodal Use
This project demonstrates multimodal AI through several complementary applications.
AI supported the design of interactive learning activities within Blackboard and H5P, generated text-based examples for students to critique, produced simulated research datasets for authentic assessment activities and generated code to support biological data analysis.
Students interacted with learning through multiple formats including structured digital forms, written AI-generated examples, simulated scientific data and reflective activities. Rather than using AI solely to produce answers, the emphasis was on analysing, evaluating and improving AI-generated outputs.
What did this look like in practice?
Students completed structured supervision logs throughout their dissertation, documenting meetings, laboratory work, agreed actions and reflections. This enabled both supervisors and the module leader to monitor engagement and intervene where necessary.
During dissertation writing workshops, students compared AI-generated examples of scientific writing with their own work, identifying why some examples demonstrated stronger academic practice than others.
Within assessment activities, students analysed AI-generated datasets before producing reports from the perspective of professional medical writers, connecting academic learning with real-world scientific practice.
The dissertation supervision log also helped meet external accreditation requirements by providing documented evidence of supervisor-student interactions.
What was the impact?
The AI-supported supervision log enabled earlier identification of students who required additional academic support while improving oversight of supervisor engagement throughout the dissertation process.
The structured approach also provided evidence to support Royal Society of Biology accreditation requirements by documenting supervision activities and student progress.
Using AI-generated examples during teaching encouraged students to think more critically about scientific writing rather than accepting AI outputs uncritically. Students were encouraged to evaluate AI-generated content, identify weaknesses and improve their own work.
Assessment redesign also increased authenticity by aligning tasks more closely with professional roles that graduates may encounter within the life sciences sector, helping students appreciate how AI can support—but not replace—expert judgement.
Although several initiatives remain under evaluation, the first implementation demonstrated promising opportunities to improve learning, student support and assessment design while encouraging responsible AI use.
Ethical or practical considerations?
Ethical AI use formed a central part of Lee’s approach.
Students were encouraged to use AI as a critical thinking partner rather than a tool for generating final assessment submissions. Reflection on AI use—including prompts, outputs and decision-making—was incorporated into assessment activities to promote transparency and responsible practice.
Particular attention was also given to data privacy when working with biological datasets. AI could be used appropriately to generate analysis code, but sensitive patient or research data should never be uploaded into external AI systems.
Broader discussions also explored sustainability, copyright, responsible prompt engineering and the importance of preparing graduates to use AI ethically within future workplaces rather than attempting to prohibit its use entirely.
Reflections and Advice
What worked well?
Using AI to design learning activities significantly reduced development time while supporting richer student engagement. The dissertation supervision log provided valuable insight into student progress, while AI-generated examples helped students develop stronger critical writing skills. Authentic assessment tasks also strengthened links between academic study and professional practice.
What would you refine or change next time?
Future work will evaluate the effectiveness of the supervision log over multiple cohorts and gather formal student feedback. Additional guidance will also be developed to encourage more reflective supervision records and improve the quality of student engagement.
Further work is planned around prompt engineering, inclusive AI use and continued assessment redesign.
What advice would you give to colleagues considering something similar?
Begin with the educational challenge rather than the technology. Use AI to enhance teaching design, encourage critical thinking and support authentic learning rather than simply generating content. Work with students to discuss appropriate, ethical and transparent AI use so they develop confidence alongside responsible practice.
Any further reflections?
Lee views AI as becoming an integral part of teaching, research and professional practice rather than a separate innovation. As AI becomes increasingly embedded within higher education, the priority should shift from asking whether students should use AI to helping them understand when, why and how to use it responsibly. Assessment design should evolve accordingly, focusing on authentic tasks where critical thinking remains central regardless of AI capability.
Quick start guide for colleagues
Estimated time required to set up: Approximately 30–60 minutes for developing AI-assisted H5P activities; longer when redesigning assessments.
Digital skill level required (Low / Moderate / High): Moderate
Common pitfalls to avoid:
- Using AI to replace critical thinking rather than support it
- Uploading confidential or sensitive research data into external AI platforms
- Focusing on AI tools instead of the underlying learning outcomes
- Designing assessments that encourage reproduction of AI-generated content rather than evaluation and application
Best suited for (discipline / module type / cohort): Dissertation modules, laboratory-based programmes, research methods, postgraduate taught courses, Life Sciences, Health Sciences, Biological Sciences, and any module involving academic writing, supervision or authentic assessment.
