Efficiency at what cost? Salvaging constructive alignment from the GenAI hype

Synthesis: McInnes, Airey, Moss, and Rathnappulige (2026) apply critical discourse analysis (CDA; Fairclough, 1995) to 14 pieces of online gray literature — institutional web pages, open-access book chapters, and blogs from universities in North America, Australia, and Europe — offering advice on using generative AI (GenAI) for constructive alignment in higher education. Drawing on critical and post-digital theory (Fawns, 2019; Stommel et al., 2020), the authors argue that this advice, though framed as practical guidance, circulates an authoritative and efficiency-saturated discourse that reframes curriculum design as a promptable procedural workflow. Rather than critiquing GenAI use per se, they contend that universalised, decontextualised prompts misappropriate Assessment and intended learning outcomes for performative compliance, devalue academic and affective labor, and reposition educational developers as technical implementers rather than pedagogical collaborators. Their remedy is not prohibition but re-sequencing: educators must first possess sufficient understanding of constructive alignment — grounded in constructivist epistemology — to direct, evaluate, and reject GenAI output, and defensible augmentation then requires institutionally bounded, retrieval-augmented systems configured around local policy, quality standards, and situated Pedagogies and Teaching Strategies.

Key Findings

• The sampled advice promotes efficiency over pedagogical depth: Across all 14 sources, pedagogical vocabulary was largely absent and the rationale for GenAI was expressed through techno-solutionist efficiency claims — describing it as "an effective and efficient way to draft rubrics" or able to "make course design more efficient, aligned and engaging" — reducing constructive alignment to task automation rather than co-construction.
• Anthropomorphic framing elevates the tool to pedagogical peer: Texts positioned GenAI as an "educational expert and assistant," a "sparring partner," or "intelligent assistants in instructional design," creating a "discursive partnership" that implies the system holds the same contextual and ethical discernment as an educator and reframes design as a socialised interaction with an algorithmic colleague.
• Academic expertise is discursively partitioned: Staff were cast as supplying subject matter expertise while GenAI handled the "heavy lifting of developing learning objectives, organizing course content, creating course outlines, and aligning course components" — an industrial metaphor that devalues the cognitive and pedagogical labor of design and casts the expert as supervisor to a more efficient automated laborer.
• Two ideological formations drove the discourse: Analysis surfaced an AI hype formation (innovation, disruption, inevitability — "GenAI is here to stay!" — and deference to GenAI's supposed objectivity over human blind spots) and an institutional pressures formation rooted in managerialist and neoliberal logics that treated constructive alignment as an auditable output to be streamlined.
• Three substantive harms were identified: Performativity — alignment that merely looks aligned, prioritizing form and formatting over context and purpose; omission of situated pedagogies — no meaningful engagement with sociocultural, disciplinary, or institutional context; and shallow constructive alignment — emphasizing the "alignment" while neglecting the "constructive," generating outcomes, activities, and assessments in isolation as discrete rather than interdependent elements.
• Educational developers were shut out of the loop: The sources framed EDs technicistically — "helping faculty acquire fluency" — repositioning them as technical trainers mediating software adoption, sidelining their relational, values-first contribution and removing them from the closed educator–GenAI loop.
• The proposed remedy is sequencing, not prohibition: Defensible augmentation requires that educators already understand constructive alignment well enough to direct, interrogate, and where necessary reject GenAI output; without that schema, surface acceptance on the basis of plausibility replaces genuine alignment — the same evaluative failure educators routinely warn students against.
• A bounded, institutionally configured alternative is proposed: The authors argue for a knowledge-grounded retrieval-augmented agent configured on institutional curriculum policy, rubrics, graduate attributes, and discipline conventions, which poses probing questions, flags misalignment, and escalates to human expertise at boundary conditions, plus a modest "liminal tutor" function that keeps the loop open toward the next human conversation.

Analysis & Argument

• Design and corpus: A purposive Google search in April 2025 combined GenAI terms with constructive-alignment terms and education/curriculum/university terms; the first 50 results were screened against inclusion criteria (publicly accessible English text-based gray literature, November 2022–April 2025, higher education practitioner audience). Eighteen sources met the initial criteria; theoretical saturation was reached at 14, after which no new themes emerged.
• Corpus composition: Four texts focused on learning outcomes, four on assessment and rubrics, and six on combined course-design processes. Five came from top-100 Times Higher Education-ranked institutions (three North American, two Australian), five from ranked-outside-top-100 universities, two from unranked institutions, two were open-access chapters, and two were educational blogs.
• Hybrid analytic strategy: Theoretical thematic analysis (Braun & Clarke, 2006) identified recurring linguistic and conceptual patterns using deductive codes from neoliberalism and CA literature ("efficiency," "standardization," "compliance") plus inductive codes ("technological anthropomorphism," "delegation of agency"), with reflexive peer-debriefing for credibility; Fairclough's (1995) three-dimensional model then analyzed micro (textual), meso (discursive practice), and macro (social practice) levels recursively.
• Meso level — who speaks: Most authors were centrally positioned third-space staff, educational specialists, or university learning and teaching units, publishing through institutional channels under reputational and managerial accountability; the interdiscursive blend of academic, technological, and managerial registers positioned the texts as solutions-focused interventions rather than sites of contestation.
• Micro level — how it is said: Declarative, authoritative voice; numbered steps and templates implying standardization and universality; passive constructions obscuring agency; and robot imagery reinforcing a technocratic framing of curriculum work as productisation rather than situated professional judgment.
• Macro level — the ideological struggle: The analysis reads the discourse as a contest between constructive alignment as relational, context-dependent commitment to student learning and constructive alignment as an administrative artifact suited to automation, with "innovation" invoked to resolve the tension — a resolution that erodes educator agency and recasts curriculum design as technical rather than pedagogical.
• The authors' central claim: The problem is not whether technology can play a role but whether educators engaging with it possess a sufficiently developed understanding of constructive alignment to direct and evaluate it; because constructive alignment is grounded in constructivist epistemology, where understanding is actively constructed through engaging with the interdependencies of outcomes, sequences, and assessments, delegating that process before developing the schema risks accepting plausible but ungrounded output.
• Consequences for assessment and Feedback: GenAI can map assessment tasks broadly to outcomes, but cannot know the depth, level, and interconnectedness of topics or how much learning each requires; when assessment timing and composition are generated without that mapping, feedback remains general rather than diagnostically precise and forward-looking.
• Two augmentative designs: First, the bounded agent whose authority is "derivative and bounded," operating within standards maintained by a teaching and learning unit, guiding thinking without supplying answers, designed to prevent silent delegation, retaining prompts and outputs internally as an institutional asset, and escalating rather than replacing professional judgment. Second, the liminal tutor — a provisional sounding board in the gaps between episodic ED consultations that helps educators articulate thinking and arrive better prepared, and which must follow, not precede, meaningful human engagement.

What this means for practice

• Instructors. Establish your own grasp of how outcomes, teaching sequences and assessments interdepend before you prompt a generative AI tool about any of them — the paper's central claim is that defensibility depends on your capacity to interrogate, and where necessary reject, what the tool returns.
• Instructors. Read polished, perfectly aligned output as a warning sign rather than a result: the sampled advice reduces alignment to automation, and shallow alignment produces outcomes, activities and assessments as discrete items instead of interdependent ones. Time savings are real but are not the measure of quality, so treat the tool as a reflective prompt and a source of critique rather than an efficiency shortcut, and guard the constructivist core: advice that emphasizes measurable, taxonomically neat outcomes while neglecting student meaning-making and learner diversity invites the surface application it claims to prevent.
• Designers. Build or buy a bounded, retrieval-augmented assistant configured on institutional policy, rubrics, graduate attributes and disciplinary exemplars, with escalation to human expertise at boundary conditions, rather than relying on generic internet-trained tools.
• Faculty developers. Refuse the framing of trainer who helps faculty gain tool fluency: in all 14 sampled texts the educational developer was positioned outside the educator–GenAI loop as a technical implementer, and teaching units lose their say when they accept that role.
• Administrators. Decide explicitly how much GenAI authority you authorize in curriculum design, and keep prompts and outputs inside institutional governance as a retained asset rather than letting design work pass through external tools under no quality or retention rule.

Limitations

• Fourteen gray-literature texts. The corpus is 14 publicly available web pages, open-access chapters and blogs, reached by theoretical saturation from the 18 sources that met inclusion criteria out of the first 50 Google results reviewed — a purposive sample, not a census of institutional guidance.
• Skewed institutional profile. Five of the 14 texts came from institutions ranked in the Times Higher Education top 100, 2 were educational blogs and 2 were open-access chapters, so the discourse analyzed may not represent how most universities advise on alignment.
• Discourse analysis, not effects. The study analyzes how advice is worded at the micro, meso and macro levels of critical discourse analysis; it examined no curriculum, no educator and no student outcome, so the claimed harms of the efficiency framing are argued rather than measured.
• The remedy is untested. The bounded institutional agent and the "liminal tutor" are design proposals — the paper reports no implementation or pilot and offers no evidence that either reduces the performativity and delegation risks it identifies.

Connected Concepts

• Generative AI
• Curriculum Design
• Assessment
• Educational Development
• Critical Pedagogy
• Teaching

Connected Articles

• From Experimentation to Integration: Embedding GenAI in Business Higher Education through the Lens of Constructive — Constructive alignment lens on GenAI integration in business higher education
• When faculty ask, 'what's the point of teaching?': GenAI as identity crisis, not skills gap — Faculty identity disruption amid GenAI adoption
• Don't believe the hype. AI myths and the need for a critical approach in higher education — Critical interrogation of AI myths in higher education
• Rewriting the Curriculum: A Systematic Review of Generative AI-Driven Pedagogical Change and Emerging Systems of Learning in Higher Education — Reimagining curriculum design and pedagogy with GenAI
• Governing generative AI in higher education: a global Delphi study on policy and practice — Delphi study on governing GenAI in higher education

Citation

McInnes, R., Airey, L., Moss, P., & Rathnappulige, S. (2026). Efficiency at what cost? Salvaging constructive alignment from the GenAI hype. Australasian Journal of Educational Technology, 42(4), 80–95.