Artificial and Emotional Intelligence: Two Key Forces for Teachers' Professional Development in an Era of Uncertainty

Synthesis: Aponte, Vargas Sánchez, Chiappe, and Martínez-Pérez argue that the debate about AI and teachers is stuck between technocentric optimism and pre-emptive rejection, and that the better question is under what relational conditions AI can genuinely support teachers' socio-emotional professional development. Their answer is a design thesis: AI is defensible when it works as relational infrastructure — expanding, protecting, and sustaining the human relationships tied to teacher well-being (Trust, mentoring, peer support, collaboration, psychological safety, reduced isolation) — and harmful when it acts as a symbolic substitute for human accompaniment. They introduce relational densification as the criterion by which AI-supported initiatives should be judged, and show how each family of AI-mediated functions (coordination, recommendation, analytics, conversational agents, and emotional AI) carries its own ethical and pedagogical risks.

Key Findings

• AI's contribution to teacher development is relational, not merely efficient. AI supports socio-emotional professional growth only when it functions as infrastructure that creates more human relationship — protected time for mentoring, peer dialogue, and communities of practice — rather than as an always-available "emotional agent."
• Relational densification is offered as an evaluative criterion, and it is deliberately distinct from relational density. Density describes the quality of professional relationships when they carry trust, co-regulation, continuity of support, and collective responsibility; densification names the deliberate process of expanding and stabilising those qualities. Frequency of contact or number of platform interactions does not count.
• Efficiency is not care, and freed time does not automatically become well-being. Automation may open a condition of possibility, but without an explicit institutional rule requiring that time saved be reinvested in mentoring, shared reflection, and peer accompaniment, "efficiency" is simply reabsorbed as increased demand.
• The unit of design should be the network, not the individual. Personalization typically starts from what one teacher needs; the authors argue AI should instead activate peer mentoring, connect teachers with complementary needs, maintain continuity between face-to-face encounters, and reduce isolation through durable micro-structures of support.
• Conversational agents carry a specific risk: the pseudo-bond. Because talking to an interface is low-friction, it can become the preferred option precisely when the difficult human work — negotiating, repairing, tolerating disagreement — is what would produce growth. The design response is to route teachers toward human support rather than to position the system as a companion.
• Affective data governance is a pedagogical condition, not a technical appendix. Separation between well-being support and managerial evaluation, data minimization, purpose limitation, genuinely voluntary participation, human oversight, and contestability are what make it safe for teachers to be vulnerable in AI-mediated spaces at all. This matters more as emotion-recognition systems face increasing regulatory caution in education and workplace settings.
• Critical AI literacy is itself a socio-emotional competence. Teachers who understand system limits, biases, and data practices are less likely to over-delegate judgment or naturalise recommendations, and better placed to use AI as a bridge to professional dialogue rather than a replacement for it. The authors add a related caution about the cultural fragility of algorithmic empathy: emotional meaning is situated, so systems validated in one context can invalidate or pathologise legitimate responses in another.

Study Design & Method

This is a conceptual and critical synthesis, not an empirical study: there are no participants, instruments, or fitted models. The authors build their argument by drawing together several literatures — research on teacher well-being, emotional intelligence, social-emotional learning, burnout, school climate, trust, collaboration, and collective efficacy; work on AI in education, AI literacy, learning analytics, social chatbots, and emotional AI; and critical scholarship on datafication, affective surveillance, algorithmic bias, and AI Governance — and then articulating them into a single theoretical proposition rather than testing it.

From that proposition they derive an operational framework: seven design conditions, each paired with the main risk it addresses, the institutional safeguard it requires, and possible outcome indicators. Relational densification is specified at three complementary levels of analysis — individual perceptions of being supported and able to seek help without stigma; dyadic and small-group quality of mentoring, peer feedback, and co-regulation; and organizational conditions such as protected collaboration time, sustained communities of practice, and formal separation between support and performance evaluation. The authors are explicit that the framework is a heuristic for design, implementation, and evaluation, and that the central causal claim — that AI-supported professional development strengthens these relational conditions — remains a hypothesis for future longitudinal, comparative, participatory, and social-network research.

Implications

• Change what gets evaluated. Teacher-development initiatives should be judged on whether they densify professional relationships, not only on personalization, platform use, or satisfaction. Personalization is not rejected; it is reoriented as a means of widening access to mentors, communities of practice, and resources.
• Write relational reinvestment into policy. Institutions need formal commitments that time saved through AI-supported processes flows back into collaboration and care, otherwise workload simply refills the gap.
• Treat governance as design. Data minimization, purpose limitation, opt-in and opt-out, access controls, retention limits, human oversight, and contestability belong in the pedagogical architecture, alongside formal separation from employment and evaluation decisions.
• Protect the boundary between emotional intelligence and AI. AI has no emotional experience, professional responsibility, or contextual accountability; its role should stay limited to coordination, access, referral, documentation, and preparation for reflective dialogue. Human oversight becomes a design requirement rather than a compliance checkbox.
• Expect downstream effects on students. When teachers have collegial trust, mentoring, and institutional support, they are better able to sustain emotionally responsive pedagogical relationships and psychologically safe classrooms; individualised, monitored well-being provision can degrade those conditions instead.
• Acknowledge the boundary conditions. AI cannot compensate for chronic overload, punitive accountability cultures, weak leadership, or the absence of protected time — the structural deficits that make Professional Development fail in the first place.

Limitations

• Conceptual, not empirical — relational densification is proposed as an evaluative criterion and testable hypothesis, not as a demonstrated causal mechanism.
• Broad scope across socio-emotional development, governance, and political economy means depth in any single area is limited.
• Evidence on emotional support from conversational agents is mixed and highly dependent on implementation conditions, as the authors themselves note.
• The proposed indicators are illustrative, not a validated measurement protocol; they vary with setting, resources, and system type.
• Emotional AI remains epistemologically fragile, and the framework offers no settled answer to how cross-cultural validity of affective inference should be established.

Connected Concepts

• Teaching — reframes what AI should and should not be allowed to do in teachers' professional lives
• Well-Being — teacher well-being is the outcome the paper is trying to protect
• Ethics — governs the ethical limits placed on symbolic substitution and affective inference
• AI Literacy — recast here as a socio-emotional competence, not just a technical skill
• Professional Development — institutional conditions for sustained, relational professional development
• Human-in-the-Loop — human oversight and contestability as safeguards for affective data
• Privacy — data minimization, consent, and retention limits in well-being systems

Connected Articles

• When faculty ask, 'what's the point of teaching?': GenAI as identity crisis, not skills gap — GenAI as identity work for faculty
• Generative AI in Higher Education: A Systematic Review of Opportunities, Challenges, and Pedagogical Innovations (2022–2025) — Systematic review of generative AI in higher education

Citation

Aponte, M. L., Vargas Sánchez, A. D., Chiappe, A., & Martínez-Pérez, S. (2026). Artificial and Emotional Intelligence: Two Key Forces for Teachers' Professional Development in an Era of Uncertainty. Frontiers in Psychology, 17, 1935683.