Technē, phronēsis, and poiesis: a narrative-humanistic framework for teaching innovation in the age of artificial intelligence

Synthesis: Abdelkareem (2026) proposes the Narrative-Humanistic Innovation Pedagogy (NHIP), a conceptual framework that treats literary and philosophical practice as a structural part of innovation education rather than a decorative supplement. It is a conceptual analysis informed by reflective teaching at the University of Kalba and reports no outcome evaluation. NHIP sequences four pillars: Empathic Immersion, which puts sustained close reading before stakeholder interviewing; Critical Interrogation, which requires teams to state the phronetic "why" before the technical "how"; Ethical Gray-Zone Reasoning, which uses literary cases such as Frankenstein to rehearse anticipation of consequences; and Narrative Prototyping, which develops a narrative pitch in parallel with a technical prototype. A ten-week illustrative module carries a generative AI activity through every pillar, from annotating AI summaries to auditing whether a tool should be used. The framework complements design thinking, design fiction, responsible innovation, value sensitive design, and AI literacy rather than replacing them, and it advances five propositions for empirical testing. The distinction worth keeping: the activities have been taught, but no student data were collected.

Key Findings

1. NHIP sequences four pillars into one course-level framework. Empathic Immersion, Critical Interrogation, Ethical Gray-Zone Reasoning, and Narrative Prototyping form an iterative sequence, not a one-pass procedure, each with a specified generative-AI activity.
2. A ten-week module allocates weeks 1–2 to Empathic Immersion, 3–4 to Critical Interrogation, 5–6 to Ethical Gray-Zone Reasoning, 7–9 to Narrative Prototyping, and week 10 to integration, an illustrative design, not an evaluated intervention.
3. Two meta-analyses supply the defensible claim about fiction and social cognition. Reading fiction is associated with small, context-sensitive gains, while the single-passage theory-of-mind effect reported by Kidd and Castano (2013) did not replicate.
4. Five propositions define what an evaluation would test: interview depth, documented alternatives, anticipated consequences, proposal coherence, and critical evaluation of AI output, under a matched-cohort or quasi-experimental design.
5. Six adjacent frameworks are positioned explicitly. NHIP is compared with design thinking, design fiction, responsible innovation, value sensitive design, AI literacy, and literary-humanistic education; its claimed contribution is integration, not a novel component.
6. The article separates four kinds of support and leaves the fourth empty: published research, theoretical propositions, informal reflective observations, and demonstrated effects of NHIP itself, of which there are none.

Why before how

The framework's philosophical core is a distinction, not a technique. Design-thinking curricula are typically organized around the question "how might we?", which in Aristotle's terms belongs to technē, the reasoned capacity to make. Aristotle separates poiesis, making whose end lies in a product beyond the activity, from praxis, acting well, and assigns phronēsis, practical wisdom about what is good for human beings, to the latter. Technē is therefore not the opposite of poiesis but the knowledge that guides it. Heidegger agrees by another route: technē belongs to bringing-forth, and what stands against poiesis is enframing, a challenging-forth that orders nature and people as standing-reserve. Abdelkareem uses the pairing to correct an earlier version that treated the two as opposites, and to warn that NHIP must not become a checklist. The classroom consequence is a required reflective statement of purpose before any solution is chosen.

The four pillars and their AI strand

Each pillar pairs an interpretive practice with a specific use of generative AI. Empathic Immersion uses structured reader-response tasks (what a character does not say, where sympathy shifts), then asks students to annotate an AI summary and AI "persona" against their own reading. Critical Interrogation runs a Socratic seminar on Aristotle and Heidegger and requires a written "why this, why now, why us" statement plus a whether-to-use audit and an AI-use log. Ethical Gray-Zone Reasoning treats literary narrative as moral rehearsal, then compares an AI-generated consequence list with the students' own. Narrative Prototyping, grounded in storytelling research and design fiction, develops a pitch alongside a working prototype and requires annotation of AI drafts for unsupported claims and manipulative rhetoric. The studio module is experiential, weighting a reflective portfolio as heavily as the final pitch.

What the evidence can and cannot carry

Fiction reading and social cognition are treated as contested: brief exposure effects failed to replicate, meta-analytic effects are small, and the pillar is designed around sustained engagement and a transferable interpretive skill rather than trait empathy. The weakness in student interviewing that Pillar One addresses is reported second-hand in a perspective article, not established systematically. The Kalba courses were routine teaching, with no instruments and no student work collected, so implementation shows only that the activities fit an ordinary schedule. Table 1 maps each pillar to research support, theoretical proposition, and informal observation. The five propositions in Section 7.1, with blinded raters and performance tasks, are the proposed route from proposal to evidence. The framework sits alongside design thinking and AI literacy as a complement, which also limits what it can claim as new.

What this means for practice

These are the author's design proposals, not tested effects.

• Instructors. Put a short literary text and structured reader-response tasks before the empathy stage, and assess interpretive moves (the unsaid, the shift in sympathy) rather than trait empathy.
• Instructors. Require a written "why this, and not something else?" statement before teams commit to a technical path, and make not using a tool an explicit option in the audit.
• Curriculum designers. Embed the material inside an existing innovation course rather than adding a humanities requirement, and budget seminar-sized sections for the two discussion-heavy pillars.
• Administrators. Treat adoption as staffing: the author proposes paired teaching between literature or philosophy faculty and design-trained colleagues.
• Assessment designers. Grade the reflective portfolio, including annotated AI outputs and the AI-use log, as evidence of the reasoning behind the prototype.

Limitations

• The article is conceptual and reflective, with no instruments, no student data, and no comparison group; the practice base is a single institution, the University of Kalba.
• The author designed the framework and taught the courses in which it was developed, a dual role the paper names as a bias risk; evaluation would need blind raters.
• Pillar One rests on contested evidence: brief-exposure effects have not replicated and meta-analytic effects are small, while the interviewing deficit it addresses is reported rather than systematically established.
• The illustrative texts are Western canon (Frankenstein, the Odyssey), so cultural transferability remains unresolved.
• The discussion-intensive seminars may not scale to large-enrollment courses, and no cost estimate is offered for the extra contact hours.
• No generative-AI model or version is named; the activities are tool-agnostic, so the guidance may date.

Citation

Abdelkareem, A. N. Y. (2026). Technē, phronēsis, and poiesis: A narrative-humanistic framework for teaching innovation in the age of artificial intelligence. Frontiers in Education, 11, Article 1993147.