Designing and securing assessment

The rapid evolution of AI tools has transformed what is possible for our students to do in assessment. The boundaries of what AI can do continue to expand with each new release. This has significant implications for the validity and design of assessment tasks. By proactively designing and securing our assessment we can ensure assessment outcomes reflect students' abilities and maintain the credibility of our programs.

Assessment design decisions

Professor Danny Liu (on behalf of TEQSA) has created short videos demonstrating AI's capabilities:

Permitted use of AI in your assessment

Each Course Coordinator must make decisions about the most appropriate use of generative AI and Machine Translation (MT) in their course and for their assessment.

When AI is permitted:

Set clear expectations: Outline how students should acknowledge or reference the tools used.
Support students to use AI ethically and effectively.
Test your assessment:
- Use Microsoft Co-pilot to simulate AI's role in completing tasks.
- Experiment with scaffolds and prompts to guide students, e.g., "Provide an outline for..." or "Identify key ideas for...".
- Regenerate AI responses to assess variability and adaptability.

Guidance on designing assessment, examples of assessment ideas and learning design support is available if you wish to make changes to your assessment. Any changes to assessment need to be made in accordance with the assessment policy and procedure with due consideration of accreditation requirements and the impact on programs.

What is secure assessment?

Secure assessments are designed to ensure confidence that the work reflects the student’s own learning. These tasks have some level of supervision and are designed to:

minimise opportunities for academic misconduct
ensure fairness in grading
make detection of misconduct highly likely.

Secure assessment strategies include supervised tasks, invigilated activities, or assessments involving direct interaction with staff or peers. Across UQ, schools and faculties are working to clarify what secure assessment looks like in their contexts.

Associate Professor Ryan Walter from the School of Politcal Science has developed a list of secure assessment approaches:

Live (in-person) quizzes in lectures
Live (in-person) quizzes in tutorials
In-class padlet submission applying concepts to cases
Mock interview focused on “live” analysis of a case presented in the interview
Oral interactive assessments (alone or added to a take-home assignment)
Interactive and in-person simulations
Practical placements (assessed by supervisor)
Supervised competency practicals, quizzes or exams
Peer-reviewed group work or assessment
Invigilated Exams

Examples of secure assessment tasks

Secure assessment can be implemented in many ways to suit the context and size of a course.

Associate Professor Peter Lewis coordinates Advancing Research Inquiry (HLTH7315) with a Research Question assessment including a viva voce component. The viva voce has been effective at assessing students' abilities and engaged students with personal, real time feedback.

Associate Professor Lewis provides students with:

The following non-exhaustive list highlights examples from the Assessment Ideas Factory of secure assessment.

Authentic assessment in Finance

500+ students
Post-graduate
High time required
Conditions: Work-related, Group, Peer-assessed

Authentic weekly portfolio, case studies, team-based assignment, and reflection

40-60, 60-80, 80-100 students
Post-graduate
Medium time required
Conditions: Group, Peer-assessed, Time limited

Clinical Project Using Action Learning in Health Sciences

80-100 students
Third year/Post-graduate
Medium time required
Conditions: Work-related, Group, Sequence

In-class writing tasks in public health

100-500 students
First year
Low time required
Conditions: Sequence

Objective Assessment of Complex Skills Using OSPE Examinations

100-500 students
Third year
High time required
Conditions: Identity verified, Work-related

Oral Exam in Political Science

120 students
Second year, Third year, Post-graduate
Low time required
Conditions: Identity verified, Work-related, Time limited

Pair-Interview (assessed conversations)

20-40 students
First year
Medium time required
Conditions: Identity verified, Group

Practical Activity Facilitation (demonstrating a competency)

80-100 students
First year, Second year, Third year, Post-graduate
High time required
Conditions: Identity verified

Unfamiliar Contexts in Invigilated Exams

500+ students
First year
High time required
Conditions: Identity verified, Time limited

Examples of course assessment profiles with secure assessment 

Examples of rubrics for assessment allowing AI 

Allowing AI in assessments shifts the focus of what is being evaluated. Rubrics should reflect these changes by assessing:

what new tasks students perform with AI (e.g., prompt creation, tool integration)
what students no longer do manually (e.g., grammar correction or basic summarisation).

Examples of rubric adjustments

Adding, removing or adapting criteria and standards related to:

grammar and expression
academic voice
depth of personal insight and reflection
use of specific rather than generic examples
connection between evidence and arguments
direct reference to class discussions and activities
acknowledgement of complexity and nuance in a topic
evidence of revision of ideas
the effective use of AI.

Example Rubrics

Research essay (qualitative rubric) (PDF, 44.9 KB)