AI Considerations for Assessment Design

Generative AI is now widely available and increasingly embedded in education and professional practice. While AI can support learning, it challenges traditional assessment approaches because students can use it to produce work that appears to demonstrate achievement without necessarily evidencing their learning.  

Rather than attempting to ban AI or relying on detection methods, programme teams should consider how assessment can provide valid evidence of learning. The key question is not, "How do we stop students using AI?" but "How do we design assessments that require students to demonstrate learning, with or without AI?"  

Acceptable AI Use  

AI can support learning when it enhances, rather than replaces, students' thinking, engagement and decision-making. Appropriate AI use will vary according to the learning outcomes, discipline, level of study and assessment requirements. 

The following principles should inform decisions about AI use in learning and assessment:

Transparency

Be explicit about acceptable AI use and disclosure requirements. Where AI use is permitted, students should be clear about how and why it has been used, in line with LJMU guidance and assessment requirements.

Accountability

Students remain responsible for work submitted for assessment.

Critical Evaluation

Students should verify, critique and evaluate AI-generated content.

Equity and Inclusion

Expectations should be fair and accessible to all learners.

Privacy and Data Protection

Students should not enter confidential, personal or sensitive information into public AI tools.  

Environmental Sustainability

Encourage purposeful and proportionate use of AI. Where AI tools are used, they should provide clear educational value and support learning outcomes rather than being used by default.

Appropriate AI use will vary according to the learning outcomes, discipline, level of study and assessment requirements. Examples of acceptable use may include: 

  • generating ideas, questions or study prompts
  • explaining concepts and supporting understanding
  • reviewing grammar, spelling and structure
  • supporting translation, accessibility and writing development
  • providing support for coding, data analysis, and problem-solving, with the student’s oversight and verification required before implementation
  • providing feedback that students use to improve their own work

AI use is unlikely to be appropriate when it:

  • replaces rather than supports learning
  • prevents a student from demonstrating their own knowledge, skills or judgement
  • fabricates evidence, references, data or sources
  • produces work that students cannot explain, justify or take responsibility for

Framework for designing assessment in an AI-enabled environment

Design Assessment at Programme Level 

Assessment should be viewed as a coherent programme of activity rather than a series of isolated tasks. A programme-level approach provides multiple opportunities for students to demonstrate learning, development and achievement over time.

Key considerations

  • Map assessments against programme outcomes to ensure coherence and reduce duplication.
  • Use a variety of assessment methods rather than relying heavily on a single assessment type.
  • Consider integrative or capstone assessments that require students to apply learning across multiple modules.
  • Balance formative, low-stakes and high-stakes assessment opportunities.
  • Identify where invigilation is genuinely needed to assure professional competence or academic standards.
  • Review assessment patterns to ensure they provide a rich picture of student achievement across the programme.

Assess Thinking and Process 

Assessment should require students to demonstrate and explain their thinking, decision-making and learning processes, rather than relying solely on evaluation of final products.

Key considerations

  • Require students to explain decisions, justify choices and evaluate alternatives
  • Include opportunities for reflection on learning and development.
  • Assess stages of work through drafts, portfolios, project logs or research records.
  • Use presentations, vivas, demonstrations or discussions where appropriate.
  • Design marking criteria that recognise critical thinking, evaluation and judgement.
  • Encourage self-assessment and peer review to support metacognitive development.

Promote Responsible and Ethical AI Use 

Students will increasingly encounter AI in academic, professional and everyday contexts. Assessment should help them learn how to use AI appropriately, critically and ethically within their discipline.

Key considerations

  • Provide clear guidance on when and how AI may be used.
  • Align expectations with disciplinary, professional and regulatory requirements.
  • Encourage students to critically evaluate AI-generated outputs.
  • Consider ethical, legal and social implications of AI use in assessment tasks.
  • Help students understand requirements for acknowledging and attributing AI use.
  • Develop students' ability to exercise judgement rather than rely uncritically on AI-generated content.

Develop Shared Understanding  

Clear expectations about assessment standards, academic integrity and AI use are best established through ongoing discussion between staff and students.  

Key considerations

  • Discuss expectations for AI use openly in teaching and assessment guidance.
  • Use exemplars to explore standards and appropriate use of AI.
  • Clarify expectations through assessment briefs, rubrics and classroom discussion.
  • Encourage students to justify and reflect on their use of AI tools.
  • Review student understanding of expectations throughout the programme.
  • Where appropriate, involve students in discussions about assessment criteria and standards.

Review Assessment Design

When reviewing assessment strategies, consider whether students are required to: 

  • explain and justify their reasoning
  • demonstrate learning processes as well as outcomes
  • apply knowledge in authentic or professional contexts
  • discuss and defend their work through dialogue
  • show development through staged or iterative assessment
  • reflect critically on their learning
  • demonstrate skills through performance or practice
  • draw upon personal experience or individual perspectives
  • critically evaluate AI outputs and explain how they have used them.

These approaches should not be viewed as a checklist to be applied to every assessment. The most appropriate approach will depend on the learning outcomes and disciplinary context.

Summary

Ultimately, the focus of assessment should be on whether students can demonstrate achievement of learning outcomes, including their understanding, judgement and critical thinking, rather than solely on whether AI has been used.