Philosophical leadership in education: Rethinking pedagogy in an AI-driven world

Synthesis: Socorro Márquez, Reyes Ortiz and Torrez Meruvia synthesize the literature on AI and education into a role claim: the educator facing AI is less a tool user than a philosophical leader, and the role has three dimensions — ethical judgment about whether a use is worth making, pedagogical guidance that keeps the learner doing the thinking, and policy translation that turns governance language into classroom rules. The argument is teleological rather than empirical: AI supplies content and speed but no ethical discernment, so the value of an AI-mediated activity depends on an educator who can say what it is for. Its contribution to practice is a three-phase lesson template and a case for building the role through coaching and mentoring, not one-off training.

Key Findings

1. Three dimensions, in order. Ethical judgment is preparation, pedagogical guidance is execution, and policy translation is governance: the educator interrogates bias before use, directs the tool at a defined intellectual job, then enforces privacy and accountability in the room.
2. Policy translation is the dimension frameworks leave undone. International documents supply vocabulary — responsibility, transparency, trust — while the operational meaning stays with the teacher, who must turn it into a prompt rule, an assessment condition or a refusal.
3. AI enhances rather than replaces, for an ethical reason. Ethical discernment is the part of teaching no system supplies, which makes teacher agency the mechanism that converts automation into learning.
4. Coaching and mentoring are the route to the role. Philosophical leadership is a disposition developed through sustained professional relationships, not a competency acquired in a training module.
5. The historical contrast turns on one variable. Earlier automation ran on static, pre-programmed logic; generative systems are autonomous and opaque, which turns pedagogy into continuous algorithmic governance.
6. Ubuntu and Confucian ethics are governance content. Relational justice, benevolence and harmony are proposed as criteria for algorithmic decisions and a corrective to individualist frameworks, not as decoration.

Judgment before the tool is opened

The ethical dimension is prior to use, and the review's worked case is a history assignment. Before students touch a language model, the educator opens a critical dialogue about algorithmic bias, Eurocentric training data and the limits of machine neutrality, so the tool is selected and framed rather than handed over. The pedagogical dimension then constrains what it may do. The model generates counter-arguments to theses students have already written, and students fact-check its logic and hunt for hallucinated detail. That inversion is the practical content of the role: the learner's argument exists first, so the tool cannot substitute for it. The vignette closes with the educator keeping personally identifiable information out of prompts, modeling compliance instead of announcing it.

Why policy translation is a teaching job

Policy translation is the dimension the frameworks leave to whoever is teaching. UNESCO, the OECD and the European Union's AI Act supply principles — human-centered governance, transparency, risk classification, accountability — but none specifies what a teacher does on Tuesday. The review's claim is that the teacher is the mediator: a rule that cannot be restated as a prompt condition, an assessment condition or an enforceable boundary has not reached the classroom. That makes classroom-level policy the place where governance either holds or evaporates, and makes the teacher's judgment about bias a governance function, not a personal preference.

The corpus behind the argument

The method is a conceptual integrative review: a configurative qualitative synthesis with document analysis, drawing on 850 records from database searching and 75 from other sources, reduced to 780 after duplicate removal, 620 excluded at screening, 160 full texts assessed, 23 excluded and 137 included. Quantitative findings from the included trials are read for pedagogical significance rather than pooled. That is the review's strength and its limit: it connects philosophy to current empirical work, and it means the central construct arrives as an interpretation. The authors' route to building the construct is coaching and mentoring — sustained dialogue about judgment, including sessions where administrators define the principles they will enforce — rather than a tool-skills module.

What this means for practice

• Instructors: Set the ethical conversation while the assignment is being set, not after: name the bias, the training-data limits and the model's non-neutrality, then constrain the tool to a job the learner has already begun and require verification of what it returns.
• Faculty developers: Build the role through coaching and mentoring relationships rather than one-off workshops. The mechanism the review names is sustained dialogue about judgment, which is a different design from a tool-skills session.
• Administrators: Fund the coaching and mentoring the review recommends and protect time for it; a framework that exists only in a strategy document asks teachers to invent the classroom version alone.
• Policymakers: Write guidance that names the classroom decisions it is meant to reach. The finding is not that frameworks are absent but that their vocabulary does not say who does what.
• Researchers: Test the construct rather than assume it: measure whether coaching or mentoring changes ethical judgment and classroom practice, and follow a cohort beyond a single term.

Limitations

• Conceptual synthesis with no primary data: nothing measures whether philosophical leadership changes learning, ethical reasoning or classroom practice.
• English and Spanish sources only; the authors note that the search yielded Western literature and that Ubuntu and Confucian traditions enter as cited proposals, not reviewed evidence.
• Publication lag: capabilities described in the reviewed literature may already differ from the models now in schools.
• Most included studies are theoretical or short-term pilots, and the lesson vignette is the authors' own construction, not a design tested with teachers or students.

Citation

Socorro Márquez, F. O., Reyes Ortiz, G. E., & Torrez Meruvia, H. (2026). Philosophical leadership in education: Rethinking pedagogy in an AI-driven world. Social Sciences & Humanities Open, 13, 102699.