A practical tool for valid assessments in the GenAI-enabled university

While GenAI offers powerful tools that can enhance learning, it also disrupts long-standing assessment practices and raises questions about integrity, validity and fairness. For educators already navigating heavy workloads and shifting regulatory expectations, the additional complexity of designing assessments in a GenAI-enabled environment can feel overwhelming.

We worked with educators to address this challenge as we navigated course development during the merger of the University of Adelaide with the University of South Australia, now Adelaide University. We identified a pressing need for practical, accessible guidance that helps educators design valid assessments in a GenAI-enabled world. We needed a flexible model that prioritises the needs of each discipline, but also empowers educators to make informed decisions, while staying aligned with regulatory requirements and institutional priorities.

Introducing the AI Assessment Flowchart

To meet this need, we developed the AI Assessment Flowchart: a practical, self-evaluation tool that helps educators evaluate the validity of their individual assignments. Rather than prescribing a single “correct” method, the flowchart guides educators through a series of reflective questions and decision points, helping them consider different aspects of their individual assignment design. 

• When GenAI makes answers cheap, assessment must value judgement
• How to co-design learning and assessments with students and GenAI
• Ask students to create assessments for industry insight

Links embedded throughout the flowchart click through to professional learning resources and examples, so that support is immediately accessible for the university staff. By transforming a complex and often overwhelming process into a visual, interactive and action-oriented tool, the flowchart empowers educators to make confident, informed decisions, while always respecting their disciplinary context.

A conversational approach to assessment design

Central to the flowchart’s effectiveness is its emphasis on conversation and reflection rather than rigid compliance. Educators are encouraged to engage with third space professionals, such as educational designers and their colleagues, to discuss potential scenarios and share strategies. This collaborative approach acknowledges that each discipline and course context works differently, allowing educators to adapt GenAI integration in ways that align with their teaching goals and student needs. The flowchart guides decision-making by promoting dialogue, while encouraging a culture of shared learning and continuous improvement.

Navigating the five key stages

The flowchart is structured around five clear, sequential steps that guide educators through assessment design. Each stage prompts reflection and decision-making, including how GenAI can enhance learning while safeguarding assessment integrity.

Step 1. Clarify the purpose of your assignment: Educators need to ensure that the skills being assessed align with the intended course learning outcomes before considering the impact of AI. The assignment type should reflect the activities learners will undertake and correspond with the outcomes we want to measure. 

Educators also need to consider whether the assignment authentically assesses the intended skills. For example, whether it is relevant to the discipline and professional practice, and whether it engages higher-order thinking.

Step 2. AI and assessment: This is a defining step in the flowchart, asking educators to decide whether GenAI should be part of the assignment. Since students are likely to use GenAI tools for tasks such as checking spelling or generating ideas, the real question is how its use should be managed. 

The decision should reflect the professional and disciplinary context, considering how GenAI is used in practice. A “no” must be carefully justified, leading to Step 3 on securing the individual assignment against inappropriate use. A “yes” requires educators to provide clear guidance on how GenAI can be used responsibly to support learning and maintain academic integrity.

Step 3. Evaluate security: If the decision in step 2 is to not allow GenAI use, this step asks educators to consider the security of their assignment. They should identify strategies to deter students from using GenAI or consider whether the assignment should be “certified”, eg, invigilated in some way, or secure. The goal is to maintain the validity of the assignment so that learning outcomes can be measured reliably.

Step 4. Vulnerability to GenAI: This step applies when the assignment allows some use of GenAI or is potentially vulnerable to it. Educators need to consider whether GenAI could complete the assignment independently and identify ways to enhance its security. If the assignment remains insecure despite these measures, they should determine whether an alternative secure assignment exists that can validly measure the same learning outcomes. For example, instead of an essay that students complete over several weeks, switch to an invigilated essay or a viva.

Step 5. Feasibility: The final step asks educators to consider whether they have the necessary resources to support the assignment’s delivery. It also prompts reflection on whether the rubric assesses students’ actual effort and learning rather than tasks that GenAI can easily complete, such as spelling or grammar checks.

The AI Assessment Flowchart provides a practical, structured approach to designing assessments in an era when GenAI is increasingly present. The flowchart guides educators through clear, reflective steps, from clarifying learning outcomes to considering GenAI use, security, vulnerability and feasibility. These steps help ensure that the individual assignments remain valid, reliable and aligned with institutional, disciplinary and professional expectations. This tool can be used to strengthen assessment design in courses and support thoughtful, responsible integration of GenAI into learning and teaching. 

Rebecca Smith is capability development manager; Paul Moss is learning design and capability manager; Sasikala Rathnappulige is learning designer and Carina Correas is manager of educational design, all at Adelaide University.

This article benefited from discussions with Edward Palmer, Walter Barbieri, Daniel Lee and Thomas Wanner, whose insights and feedback helped shape our approach.