Assessment Design for the AI Era
AI has made some assessment questions more urgent, but many of the underlying issues are not new.
Universities are having to think carefully about what assessment is meant to evidence, how students can show their own judgement, where AI use is appropriate, how feedback supports learning, and how departments can maintain academic standards without creating unmanageable workload.
This work supports departments, programme teams and central education teams to review and refine assessment for the AI era.
The aim is not to make every assessment “AI-proof”. That language can be misleading. The stronger aim is to design assessment that is clearer about purpose, more transparent about process, more appropriate for the learning outcomes, and more workable for staff and students.
Why this matters
AI has made it easier for students to generate text, ideas, summaries, code, images and feedback. That changes the evidence universities can rely on when judging learning.
Some tasks may still be appropriate with clearer guidance. Some may need staged process evidence. Some may need to include responsible AI use. Some may need more secure checkpoints. Some may need to become more authentic, project-based, oral, portfolio-based, practical or reflective. Some traditional essays or exams may still be the right choice.
The work is not to replace one default with another.
The work is to understand what each assessment is trying to evidence and whether the design, workflow and guidance support that purpose.
Who this is for
This support is useful for:
-
departments reviewing assessment in response to AI;
-
programme teams wanting a more coherent assessment mix;
-
academic leads balancing authenticity, rigour, workload and standards;
-
digital education or academic development teams supporting staff with AI-era assessment;
-
quality, assessment or registry teams needing clearer links between design, process and assurance;
-
institutions trying to move beyond generic AI guidance into course-specific assessment practice.
What this can include
Support can include:
-
reviewing current assessment tasks and assessment patterns;
-
mapping what each task is intended to evidence;
-
identifying where AI changes the risk, purpose or evidence base;
-
refining assessment briefs, criteria and student guidance;
-
supporting more authentic, project-based, staged, oral, portfolio or process-based assessment where appropriate;
-
identifying where secure checkpoints or individual verification may still be needed;
-
considering the balance between assessment of learning and assessment for learning;
-
reviewing feedback opportunities and how students use feedback;
-
aligning assessment design with Moodle/platform workflows;
-
considering workload, moderation, external review and quality assurance implications;
-
facilitating departmental workshops to develop practical options.
A balanced approach
AI-era assessment design should not become panic redesign.
Authentic assessment can be powerful, but it is not automatically better. Project-based work can be meaningful, but it can also create workload, moderation and equity challenges if it is not designed carefully. Oral assessment can support verification, but it needs clear criteria, scheduling, consistency and accessibility considerations. Process evidence can help, but it needs careful interpretation.
The strongest assessment designs tend to balance several needs:
-
meaningful learning;
-
trustworthy evidence;
-
academic standards;
-
student clarity;
-
staff workload;
-
inclusive practice;
-
platform/workflow fit;
-
manageable moderation and quality processes.
Typical outputs
Depending on scope, outputs can include:
-
assessment design review;
-
assessment purpose/evidence map;
-
AI-era assessment risk and opportunity summary;
-
redesigned assessment brief options;
-
departmental workshop materials;
-
student guidance recommendations;
-
Moodle/platform workflow implications;
-
feedback and process-evidence recommendations;
-
programme-level assessment pattern review;
-
pilot or implementation recommendations.
When this fits
This is a good fit when a department or programme knows assessment needs attention, but does not yet want to jump straight to a platform decision or full workflow redesign.
It is also useful when AI guidance exists at institutional level, but departments need help translating that guidance into assessment tasks, student instructions, feedback practice and workable local processes.
Related support
See also:
Frequently asked questions
Does AI mean essays and exams are no longer useful?
No. Essays and exams can still be useful when they are the right fit for the learning outcomes and evidence needed. The work is not to replace one default with another, but to understand what each assessment is trying to evidence.
​
What is AI-era assessment design?
AI-era assessment design reviews how assessment tasks, criteria, feedback, student guidance and evidence of learning need to adapt in a context where students and staff may use AI tools.
Does authentic assessment solve academic integrity concerns?
Not automatically. Authentic, project-based, staged, oral or portfolio assessment can be powerful, but it still needs clear criteria, fair workload, manageable moderation and careful guidance.
​
Why connect assessment design with workflow?
A redesigned assessment still needs to work in practice. Moodle/platform setup, marking, moderation, feedback, student communication and quality processes all affect whether the assessment is manageable and trustworthy.