Ethical principles of AI in education: Exploring teachers' contextual ethical reasoning through an STS lens

Synthesis: Adelana, Ebubedike, Crabb, and Rienties (2026) examined how in-service secondary STEAM teachers in a low-resource Nigerian context understand the Ethical Principles of AI in Education (EPAI-Ed). Under a Socio-Technical Systems (STS) lens they ran a two-stage qualitative design: a 21-item think-aloud protocol (n = 30) and follow-up interviews (n = 10). Despite lacking formal AI ethics training, teachers are not "ethics empty" — they reason from common sense and professional experience, shaped by professional identity, classroom realities, cultural norms, and contextual vulnerabilities. They converged on transparency, human oversight, and co-designed governance but split over context-specific training data, and the authors argue for aligning their context-based beliefs with global frameworks.

Key Findings

1. Teachers are not "ethics empty." Thirty in-service secondary STEAM teachers in a Southwest state of Nigeria (20 male, 10 female) reasoned substantively about AI ethics without formal training, and seven of the ten interviewed reported minimal-to-no prior exposure to formal AI ethics principles.
2. Context and vulnerability drive ethical reasoning. Judgments were shaped by professional identity, classroom realities, cultural norms, and contextual vulnerabilities such as funding constraints, policy gaps, and infrastructural deficits — not by abstract principle lists.
3. A two-stage qualitative design. A 21-item think-aloud protocol on six principles (privacy, autonomy, transparency, responsibility, fairness, trust) was analyzed with sentiment coding (Cohen's Kappa between 0.629 and 0.802) and reflexive thematic analysis of the interviews in NVivo 15.
4. Near-consensus on agency, transparency, and accountability. 28 of 30 teachers (94%) wanted users to hold the final say over AI outputs, 26 (86%) rejected significant AI decisions without human oversight, all 30 demanded understandable AI decision logic, 28 (94%) wanted explicit accountability guidelines shared with providers, and all 30 wanted teachers involved in AIED development.
5. Fault lines and named risks. Teachers split over country-specific training data (14 for, 11 against, 5 undecided) and over teachers supporting AI (9 for, 3 against, 18 undecided, favoring collaboration); interviews surfaced student misuse (seven), over-reliance (seven), academic dishonesty (six), threats to teacher roles (five), and social isolation (three). All ten interviewees wanted context-sensitive policy that teachers help design.

Ethics as Situated, Not Abstract

The paper rejects the assumption that ethical AI use flows from learning ethical principle lists. Its design — a think-aloud protocol read through Socio-Technical Systems theory, then interviews — watches teachers reason rather than recall rules. Ethical awareness is situated: produced in the interaction between professional identity, classroom demands, and the constraints of a low-resource setting. Teachers "do" ethics from experience and cultural norms, so an abstract, one-size-fits-all EPAI-Ed framework misses how responsible AIED is practiced; in STS terms their ethical vigilance compensates for a weak technical subsystem.

Agreement, Division, and Trust

Agreement ran high: all 30 teachers wanted AI decision logic explained, 28 (94%) wanted users to retain the final decision, 25 (83%) denied that AI poses no risk to collected personal data, 25 (84%) rejected normalizing favoritism in AIED outcomes, and 28 (94%) rejected out-of-context training data. The sharpest split was whether teachers should support AI, where the undecided majority favored a collaborative model in which the teacher decides whether to accept AI suggestions. Trust emerged as the integrative value binding the other principles, earned by understandable, contextually aligned, co-designed systems rather than technical sophistication.

The Low-Resource Gap in AI Ethics Research

The study's second contribution is empirical: the authors note little field evidence from the Global South on how teachers in low-resource contexts conceptualize AI ethics, and the six principles come from Jobin et al.'s (2019) global mapping because they are well established yet seldom tested in this setting. The STS framing ties teachers' reasoning to the socio-technical constraints of their schools — limited-to-nonexistent AIED regulation, fragile data protection practices, infrastructural deficits, and no formal AI ethics training — so responsible AI use cannot be separated from questions of infrastructure, resources, and power, and policies exported from richer settings must be reinterpreted locally.

What this means for practice

• Instructors. Build AI ethics professional development from teachers' existing professional reasoning rather than abstract principle lists.
• Teacher educators. Use think-aloud and case-based discussion, then connect local dilemmas to formal EPAI-Ed principles to avoid superficial "checklist ethics."
• Policymakers. Co-design AIED governance with teachers rather than importing global frameworks wholesale, prioritizing data protection, transparency, and clear accountability.

Limitations

• These are baseline findings in an ongoing multi-phased longitudinal study: thirty STEAM teachers from a single state in Nigeria limit generalizability to other regions and to other professional and educational backgrounds.
• Teachers' accounts are self-reported and may reflect aspirational rather than fully enacted practice, with social desirability bias possible on universally recognized principles such as privacy, responsibility, and fairness.
• AI technologies, policies, and discourse evolve rapidly, so the findings capture teachers' viewpoints only as of data collection (November 2024 – January 2025).

Connected Concepts

• Ethics
• Global South
• Equity
• AI in Education
• Teaching
• AI Literacy

Connected Articles

• Situated AI ethics: a cultural-historical and ecological framework for education — Situated AI ethics: a cultural-historical and ecological framework
• Understanding ethical dimensions of AI in higher education: insights from faculty members and students — Understanding ethical dimensions of AI in higher education
• Beyond Compliance: A Proposed Framework for Ethical Governance of Student Data in Learning Analytics — Ethical governance of student data in learning analytics
• Generative Artificial Intelligence Policy: A Qualitative UNESCO Framework Analysis — Generative AI policy: a qualitative UNESCO framework analysis

Citation

Adelana, O. P., Ebubedike, M., Crabb, E., & Rienties, B. (2026). Ethical principles of AI in education: Exploring teachers' contextual ethical reasoning through an STS lens. Computers and Education Open, 100423.