Designing assessments in the generative AI era: A tailored assessment framework for ICT tertiary education

Synthesis: Ahangama (2026) confronts the gap between the speed of generative AI adoption and the pace of assessment reform in ICT tertiary programs. Using a qualitative interpretivist design and thematic analysis of 80 publicly available scholarly and institutional documents, the study derives seven themes and 70 subthemes and converts them into two complementary models. GENESIS (GenAI-Enabled Strategic and Instructional System for Assessment) is a five-pillar, institution-wide framework whose first four pillars, Values and Dispositions, Governance, Support Infrastructure, and AI-Integrated Curriculum, form a sequential pathway of institutional readiness. The fifth pillar, Targeted Assessment Design, is operationalized through the Assessment Design Matrix (ADM), a tripartite structure linking graduate competencies, assessment strategies, and assessment methods so that competencies inform strategy selection, strategies guide method choice, and methods generate valid evidence of attainment. The core claim is that valid assessment in the GenAI era is not a technical fix but a whole-of-institution transformation that discipline-tailored design tools can turn into concrete assessment decisions.

Key Findings

• GENESIS is a five-pillar, institution-wide framework: Values and Dispositions, Governance, Support Infrastructure, AI-Integrated Curriculum, and Targeted Assessment Design.
• The first four pillars are sequential: shared institutional values underpin governance structures, governance enables support infrastructure, and that infrastructure allows AI to be integrated into the curriculum.
• Pillar five is the Assessment Design Matrix (ADM), a tripartite structure in which competencies inform assessment strategies, strategies guide method choice, and methods produce valid evidence of attainment.
• The ADM catalogues 10 graduate competencies (GC1 to GC10), including prompt engineering, metacognitive skills, ethical intelligence, teamwork, and GenAI literacy and strategic use.
• The ADM catalogues 18 assessment strategies (AS1 to AS18) and 19 methods (AM1 to AM19), ranging from in-person foundational knowledge checks to hackathons, process portfolios, simulations, and capstone projects.
• Thematic analysis of 80 documents from 2023 to 2025 yielded 7 themes and 70 subthemes from 185 initial codes, with saturation reached after 72 documents, and six sample ICT assessment briefs illustrate the matrix in use.

The GENESIS framework

The seven themes were re-ordered into a five-pillar framework. GENESIS opens with Values and Dispositions, arguing that a culture of innovation, a facilitative rather than policing stance toward GenAI, and shared responsibility must exist before policy or tooling changes hold. Governance follows, translating those values into formal policy: integrity rules, quantified allowances for AI use, misconduct detection that avoids the false-positive harms reported for non-native English speakers, privacy safeguards, and equitable access. Support Infrastructure is third, covering staff upskilling, automated assessment systems, resource equity, and AI-enabled tutoring support. AI-Integrated Curriculum is fourth, embedding GenAI literacy as a graduate attribute and aligning assessment across whole programs. Each stage supplies the precondition for the next, and Targeted Assessment Design, the fifth pillar, is the culminating enactment of the earlier reforms. The author presents the model as complementary to Australia's TEQSA national implementation toolkit.

The Assessment Design Matrix

The ADM is offered as a practical extension of pillar five. It links three catalogued dimensions through reciprocal relationships: design-level alignment between graduate competencies and assessment strategies, a design-to-implementation bridge from strategies to methods, and implementation-level alignment verifying that methods yield valid evidence of the targeted competencies. At least one element from each dimension must be selected, and designers iterate until alignment is defensible. A worked example from an IT project management unit targets teamwork and conflict resolution: the strategy "assess process over product" is enacted through a process portfolio documenting team charters, recorded conflicts and resolutions, decision records, individual reflections, and meeting summaries. The paper maps the matrix against constructive alignment and evidence-centered design, arguing that it adds graduate competencies, industry relevance, and a verification step while spanning both formative and summative purposes.

Method, evidence, and positioning

The study uses web-based document analysis with Braun and Clarke's six-phase thematic analysis, a design the author defends as suited to synthesizing an emerging body of gray and published material. The dataset comprises 80 publicly available documents from 2023 to 2025 gathered through structured searches in March 2025, excluding sponsored links and anything lacking professional affiliation. Coding ran from April to May 2025 and produced 185 initial codes. Two LLMs were used for copy-editing only. Existing sector responses, including traffic light systems, the two-lane approach, and the AI Assessment Scale, are described as valuable but generic, focused on communicating rules, and unvalidated in authentic settings, with GENESIS and ADM offered as the discipline-tailored and systemic alternative.

What this means for practice

• Assessment designers. Start from the competency rather than the task: name the target competencies, select strategies that can validly measure them, then choose methods, checking alignment in both directions.
• Instructors in ICT programs. Favor methods that leave process evidence, such as portfolios, debugging and code tracing, milestone group projects, and capstone work, since single-point artifacts are what GenAI produces with minimal engagement.
• Program and policy leaders. Values, governance, and support infrastructure are preconditions, so course-level redesign without policy clarity and staff development is likely to stall.
• Institutional policy teams. Build the facilitative route alongside the rules, since the analysis links punitive detection regimes to inequitable outcomes for non-native English speakers.

Limitations

The models are conceptual and were not empirically validated; the author calls for pilots and longitudinal, mixed-methods testing. The dataset was limited to 80 publicly accessible, predominantly English-language documents published in a short window, so internal practices and tacit institutional knowledge are absent. The analysis involved a single researcher, with no inter-coder reliability check reported. The framework explicitly targets ICT programs at AQF levels 7 to 9 in Australia, and transferability elsewhere is asserted rather than demonstrated. Implementation challenges such as staff resistance, resource limits, and workload pressures are acknowledged but not resolved.

Connected Concepts

• Assessment Validity: valid evidence that students attained the intended competencies
• Authentic Assessment: process-oriented, integrity-focused design
• Generative AI
• Academic Integrity
• AI Governance
• Curriculum Design
• Higher Education

Connected Articles

• A bit of chaos and madness: The AI Assessment Scale and the work of assessment reform: The AI Assessment Scale and assessment reform
• Beyond Detection: Redesigning Authentic Assessment in an AI-Mediated World: Beyond Detection, authentic assessment redesign
• Coauthorship integrity: Reconceptualising assessment validity for the age of generative artificial intelligence: Coauthorship integrity and assessment validity

Citation

Ahangama, N. (2026). Designing assessments in the generative AI era: A tailored assessment framework for ICT tertiary education. International Journal of Educational Technology in Higher Education, 23(9).