Generative AI in Learning and Teaching

What this page is for: Light-touch, practical guidance for Cardiff University staff on how GenAI can be approached within teaching, learning and assessment – particularly when designing or reviewing assessments.

 

Generative artificial intelligence (GenAI) tools are increasingly shaping how students learn, study and produce university work. Many Cardiff University students are already using these tools, and staff are encountering their impact across teaching, assessment and feedback.

This page provides light‑touch, practical guidance for Cardiff University staff on how GenAI can be approached within learning, teaching and assessment. It is designed to support colleagues in making pedagogically informed decisions, particularly when designing or reviewing assessments.

This guidance:

• Complements (but does not replace) institutional policy and intranet guidance;
• Focuses on enhancement rather than enforcement;
• Supports staff across disciplines and levels of experience.

 

 

 

On This Page

Cardiff University’s Position on GenAI

GenAI Capabilities and Limitations

Using GenAI in Teaching Practice

GenAI and Assessment Design

Supporting Students in Their Use of GenAI

Quick Reminders and Further Support

Cardiff University Approved Tools

For Cardiff staff and students, Microsoft Copilot, Google Gemini and NotebookLM are institutionally available GenAI tools. University accounts should be used to access these platforms, as they align with Cardiff University’s institutional security standards, handle data appropriately, and are approved for use within Cardiff’s digital tool environment.

---

Cardiff University’s Position on GenAI

Cardiff University’s current institutional guidance on artificial intelligence in education is available on the staff intranet and should always be the primary point of reference.

While formal University guidance on generative AI in learning, teaching and assessment continues to evolve, colleagues are encouraged to adopt transparent, proportionate and pedagogically grounded approaches that align with emerging sector practice and Russell Group principles.

Russell Group Principles (expand to read)

Cardiff’s approach aligns with Russell Group principles, which are:

1. Universities will support students and staff to become AI-literate.
2. Staff should be equipped to support students to use generative AI tools effectively and appropriately in their learning experience.
3. Universities will adapt teaching and assessment to incorporate the ethical use of generative AI and support equal access.
4. Universities will ensure academic rigour and integrity is upheld.
5. Universities will work collaboratively to share best practices as the technology and its application in education evolves.

Framing GenAI use: a two‑lane approach

In the absence of Cardiff‑specific assessment classifications, the University is currently drawing on the University of Sydney’s two‑lane assessment model as a helpful framing device when communicating expectations to students.

In brief:

Lane 1

Secure assurance of learning

• Trustworthy judgements about progress and attainment.
• Integrity of the award and individual learning protected.
• Controlled conditions, contextual task design, or verification.

Lane 2

Open assessment for and as learning

• Open, supported assessment outside controlled settings.
• AI use cannot be prohibited in unsecured contexts.
• Focus on ethical, critical and discipline-specific use.

 

The two-lane model should not be interpreted as permission to ban or permit GenAI universally across a module or programme. Instead, it supports balanced, outcome-led decision-making. Staff should consider which learning outcomes require students to demonstrate knowledge, skills or judgement independently, and where these may need to be assessed in a secure or verifiable context. Equally, where GenAI use is appropriate within open assessment contexts, students should be supported to use it critically, transparently and in ways that align with the intended learning outcomes. The focus should be on what needs to be evidenced, not on applying blanket rules to whole assessments.

↑ Back to top

---

GenAI Capabilities and Limitations

GenAI tools are developing rapidly. The points below reflect typical capabilities and limitations at the time of writing and should be interpreted cautiously as tools continue to evolve.

GenAI tools can support learning in many ways, particularly when used to explore, question and develop understanding rather than simply produce finished work. For example, students and staff may use GenAI to:

• Generate ideas, questions or initial outlines;
• Explore unfamiliar topics or receive alternative explanations;
• Summarise, rephrase or compare concepts;
• Test understanding through dialogue, prompts or practice questions;
• Support coding, debugging or problem-solving;
• Draft, redraft or structure arguments;
• Produce examples, scenarios or teaching explanations.

As with any source of information, AI-generated responses should be checked, questioned and contrasted with reliable academic, professional or disciplinary sources.

Key limitations and risks

GenAI tools:

• Produce probabilistic outputs, not verified knowledge;
• May generate confident but incorrect information (“hallucinations”);
• Can fabricate references or misrepresent sources;
• Lack contextual and disciplinary judgement;
• May reinforce bias or exclude marginal perspectives;
• Often produce fluent output that masks surface‑level understanding.

Academic considerations

Generative AI tools present significant opportunities for teaching and learning, but also introduce important challenges for assessment. This guidance outlines key risks identified in recent research and highlights implications for practice.

↑ Back to top

---

Using GenAI in Teaching Practice

When used thoughtfully, GenAI can support aspects of teaching preparation and student learning without replacing academic expertise or judgment.

Examples of constructive use

Staff may choose to use GenAI to:

• Generate varied explanations or examples for lectures;
• Produce practice questions or formative activities;
• Draft or refine marking criteria and rubrics;
• Create feedback prompts that support student revision;
• Offer alternative explanations or summaries to support accessibility.

Modelling responsible use

Good practice includes:

• Being open with students about how and why you use AI tools;
• Avoiding full automation of teaching, marking or feedback;
• Critically reviewing and editing all AI‑assisted outputs;
• Framing AI as a support for learning, not a substitute for thinking.

↑ Back to top

---

GenAI and Assessment Design

GenAI has significant implications for assessment, particularly in open coursework contexts. Rather than attempting to make assessments “AI‑proof”, the focus should be on alignment with learning outcomes, authenticity and clarity.

Effective assessment design asks:

Assessment Design Menu: GenAI considerations by assessment type

Decisions about GenAI use are most effective when considered coherently at programme level, rather than independently for each module or assessment. Staff should begin by reviewing the learning outcomes that need to be assessed, then consider which assessment methods best enable students to evidence those outcomes, including how risks of misconduct or unethical AI use will be managed. The Assessment Compendium provides examples of common assessment types used at Cardiff and may help staff identify suitable methods and design features.

For each assessment, staff are encouraged to consider:

• Which learning outcomes, knowledge, skills or judgement the assessment is intended to evidence;
• Which assessment methods or design features best enable those outcomes to be assessed;
• What risks to validity, integrity or equity need to be considered, including unethical or undeclared use of GenAI;
• How expectations about GenAI use, acknowledgement and evidence of learning will be communicated to students;
• Where GenAI is available or permitted, how the assessment will evidence learning beyond the polished final output.

Where possible, expectations around GenAI use should be discussed and aligned at programme level. This helps students understand how different assessments develop complementary skills and why expectations may vary across modules, rather than encountering apparently inconsistent rules.

↑ Back to top

---

Supporting Students in Their Use of GenAI

Students are likely to use GenAI tools regardless of whether they are explicitly discussed. Clear, consistent guidance supports academic integrity, inclusion and digital capability development.

Talking points for staff

When discussing GenAI with students, consider:

• Explaining whether an assessment sits in a secure or open context, and why;
• Being explicit about what kinds of GenAI use are suitable/recommended/suggested for the task.
• Emphasising the importance of critical evaluation of AI output;
• Reinforcing that GenAI should support, not replace, learning and thinking.

Where appropriate, expectations should be reflected in:

• Assessment briefs;
• Marking criteria;
• Worked examples or exemplars;
• Teaching and learning activities.

In most cases, students should be encouraged to reference original sources rather than the AI tool itself. Expectations around acknowledgement should be proportionate and aligned with the purpose of the assessment.

Student‑facing guidance on GenAI use is available via the Cardiff intranet and should be signposted consistently.

↑ Back to top

---

Quick Reminders and Further Support

Key points to remember

If you are unsure how GenAI intersects with your assessment or teaching context, the Learning and Teaching Academy (LTAcademy@cardiff.ac.uk) can support you with programme-level conversations, assessment design, and digital education advice.

Where Next?

↑ Back to top

---