The Policy Deficit in AI × Social-Emotional Learning Research

Synthesis: A systematic review of 65 peer-reviewed papers at the intersection of AI and social-emotional learning (SEL) finds a substantial "policy deficit": nearly three-quarters of studies make no mention of policy implications at all, and those that do often lack the actor-specific detail needed for evidence-informed policymaking. Using a "WH-question" framework (Who, What, Why, When/Where, How), the authors show that policy engagement correlates with publication venue, reflecting academic incentives that favor technical novelty over AI Governance and AI Regulation in Education. They warn of a "techno-solutionist" trap and call for a shift from "implication-as-afterthought" to "implication-as-methodology," offering guidelines for researchers, editors, reviewers, and policymakers.

Policy deficit in AI × SEL research

As generative AI and emotion-aware systems are increasingly integrated into social-emotional learning initiatives, the need for evidence-based policy has grown. This systematic review of 65 peer-reviewed papers examined how studies at the AI–SEL intersection articulate policy implications.

A majority of studies ignore policy

The central finding is a policy deficit: nearly three-quarters of the reviewed studies did not mention policy implications at all. Only about one in four offered any policy recommendation, and few provided detailed, actor-specific guidance on issues such as Privacy, teacher training, and resource allocation.

The WH-question framework

For the studies that did engage with policy, the authors mapped the narratives using a "WH-question" framework:

1. Who — which actors (policymakers, institutions, teachers, developers) are called to act
2. What — which specific actions are recommended
3. Why — the rationale connecting the finding to the recommendation
4. When/Where — the contexts and timing of application
5. How — the strength and framing of the recommendation

The review found that most policy narratives lacked this specificity and actor-orientation, making them weak inputs to real policymaking.

Venue effects and the techno-solutionist trap

A significant association emerged between publication venue and policy engagement, suggesting that current academic incentive structures prioritize technical innovation and pedagogical feasibility over explicit engagement with governance and regulation. This produces what the authors call a techno-solutionist trap: technical potential is foregrounded while the institutional conditions for responsible implementation remain under-specified.

From implication-as-afterthought to implication-as-methodology

Rather than treating policy as a generic ethical horizon, the authors argue AI–SEL studies should systematically specify who should act, what actions are recommended, why, when and where they apply, and how strongly they are framed. They propose a culture shift — from appending implications at the end of a paper to building implication-articulation into the research methodology itself — and offer actionable guidelines for researchers, editors, reviewers, and policymakers to bridge the gap between AI innovation and educational governance.

What this means for practice

• Researchers. Build implication articulation into the research process itself, specifying for each finding who should act, what action is recommended, why, in which context, and how strongly it is framed.
• Administrators. Require actor-specific implications in review, funding, and publication processes — of 65 reviewed papers, only 15 stated explicit policy implications and a further three contained implicit ones.
• Researchers. Specify the institutional safeguards a tool needs — Privacy, teacher preparation, resource equity — before framing AI-for-SEL deployment as ready for practice, to avoid the techno-solutionist trap.
• Administrators. Treat venue and incentive structure as levers: policy engagement correlated with publication venue, so journal and program expectations can materially change whether research informs AI Governance.

Limitations

• The search covered Scopus and Web of Science only, and non-English records, errata notes, and one retraction were excluded, so the 65-paper sample is an English-language, peer-reviewed corpus.
• Only empirical studies, meta-analyses, and systematic reviews were eligible; conceptual papers, commentaries, reports, and working papers were excluded.
• Coding of implicit policy implications is subjective, as the authors acknowledge — two authors coded individually and resolved disagreement with a third, with no inter-coder reliability statistic reported.
• No time-frame limit was applied, and each distinct implication statement rather than each paper was the unit of analysis, so papers contribute unequally to the counts.

Citation

Tran, V. C., Liu, Y., & Nguyen, V. T. (2026). The Policy Deficit in AI × Social-Emotional Learning Research. arXiv:2608.29950.