The Learning Through Assessment Framework
The University’s Learning Through Assessment Framework provides guidance on designing assessment that is meaningful, iterative, programmatic and inclusive. Assessment aligned with this framework reduces the risk of GenAI misuse by creating tasks where authentic student engagement is integral to success.
Meaningful assessment
Assessment becomes more meaningful when students understand its purpose and can connect it to their learning goals. Approaches include:
- developing marking criteria in dialogue with students to ensure shared understanding of expectations
- designing authentic tasks that reflect real-world applications of disciplinary knowledge
- offering scaffolded choices that allow students to demonstrate learning in ways that suit their strengths
- creating clear alignment between assessment tasks and intended learning outcomes.
How does this promote integrity?
Clear expectations help students understand what they are being asked to demonstrate. When students see the value in an assessment and feel trusted to engage with it authentically, they are more likely to invest genuine effort.
Iterative assessment
Iterative approaches build assessment across time, allowing students to develop and refine their work through feedback. This includes:
- portfolios that accumulate evidence of learning throughout a course
- draft submissions with formative feedback before final submission
- staged assessments where later work builds explicitly on earlier submissions
- regular low-stakes opportunities to practise and improve.
How does this promote integrity?
When staff can observe students’ work developing over time, they can gain familiarity with each student’s capabilities, interests, and voice. This can make incongruent submissions easier to identify.
Programmatic assessment
Programmatic assessment considers how individual tasks contribute to a coherent programme of study. This might involve:
- helping students identify connections between courses and design work that demonstrates integrated learning
- creating assessments that draw on students’ individual goals, backgrounds, and disciplinary interests
- designing tasks that require synthesis of material from multiple sources, including personal experience.
How does this promote integrity?
GenAI tools are less effective at producing genuinely personalised work that reflects an individual student’s unique trajectory through a programme. Tasks requiring students to synthesise their particular experiences, interests, and prior learning are inherently more resistant to generic AI-generated responses.
Inclusive Assessment
AI tools can support more inclusive assessment practices by reducing barriers that are incidental to the skills being assessed. Approaches include:
- using AI-enabled speech-to-text or text-to-speech tools to support students with specific learning differences in producing or engaging with written work
- allowing non-native English speakers to use AI to refine grammar and expression, so that language proficiency does not obscure disciplinary understanding
- incorporating AI-assisted planning or structuring tools that scaffold executive function, benefiting neurodivergent students in particular
- designing assessment briefs that are explicit about which AI tools are permitted and how, so that all students – not only those with existing digital confidence – can engage on equal terms.
How does this promote integrity?
Inclusive assessment design reduces the incentive to use AI covertly as a workaround for unaddressed barriers. When students are supported openly, they are less likely to seek undisclosed assistance. However, we should remain attentive to the equity implications of AI access itself: students vary in the tools available to them, a number of the tools and technologies have costs associated, students’ digital literacies will vary substantially (and we must not assume that our students are ‘digital natives’), and students’ confidence in using unfamiliar technologies likewise varies. These dimensions often can track along socioeconomic lines. Genuine inclusivity requires not only permitting AI use where appropriate, but actively ensuring that the support and guidance surrounding it are accessible to all.