From Abstract Ethics to Situated Practice: A Bibliometric Analysis of AI Ethics and Professional Judgement

Synthesis: Mazlan et al. (2026) conduct a descriptive–evaluative bibliometric analysis of 282 Scopus-indexed peer-reviewed articles (2009–2025) to map how AI Ethics research has shifted from abstract normative principles toward situated, practice-based ethical reasoning. Using biblioMagika® and VOSviewer, they document a pronounced post-2021 expansion of AI ethics research and a disciplinary shift toward applied domains (social sciences, education, healthcare, management). Keyword co-occurrence networks show AI ethics increasingly framed around professional judgment, Trust, Human AI Collaboration, and interpretive practice rather than technical compliance or AI Regulation in Education. Healthcare dominates citation impact, but education emerges as a conceptually important context where questions of intelligence, Learner Agency, and professional responsibility are actively negotiated. Highly cited work consistently foregrounds human-in-the-loop decision-making and contextual reasoning, supporting a practice-based, philosophically grounded reorientation of AI ethics.

Core Finding

AI ethics research is undergoing a conceptual reorientation from abstract normative principles through applied governance toward situated, practice-based ethical reasoning grounded in professional judgment. The field is no longer anchored in technical compliance or regulatory abstraction: across applied sectors, ethical reasoning is increasingly embedded in professional activity, institutional settings, and human–AI interaction.

Growth and disciplinary expansion

Publication activity was low and fragmented before 2019, rose from 2021, and accelerated sharply after 2023 (nonlinear growth, R² = 0.7949) — coinciding with widespread deployment of generative AI systems. This expansion is not confined to technical disciplines. Subject-area analysis shows Social Sciences (52.84%) dominate, followed by Computer Science (34.04%), Arts and Humanities (16.67%), and Business/Management (12.41%), with health fields (Medicine, Nursing, Health Professions) also substantial. AI ethics has matured into an applied, interdisciplinary research domain rather than a primarily philosophical or technical subfield, with ethical issues addressed as they arise within professional practice, organizational decision-making, and institutional cultures.

From abstract principles to situated practice

The study's keyword co-occurrence network reveals that terms such as professional judgment, human–AI collaboration, Trust, and decision-making occupy prominent positions close to the core "artificial intelligence" and "ethics" nodes, while purely regulatory or policy-oriented terms are less central. Interpreted through the paper's three-orientation framework — normative ethics, applied governance, and practice-based ethics — the network shows a layering: fairness and responsible AI reflect normative concerns, human–AI collaboration and education reflect applied governance, and reflective practice, reflexivity, qualitative research, and higher education point to an emerging practice-based orientation. Ethical reasoning is increasingly framed as an ongoing social and professional process rather than a problem solvable through technical optimization.

Human judgment, human-in-the-loop, and ethical agency

A unifying theme across the most influential publications is the reassertion of human professional judgment. In education, Sperling et al. (2024) conceptualize AI literacy as encompassing ethical and contextual judgment (phronesis), not merely technical proficiency; phenomenographic studies (Yau et al., 2023) show educators negotiating ethical tensions through reflective and interpretive reasoning. In healthcare and auditing, Hassan and El-Ashry (2024) link ethical AI leadership in critical care to human oversight, while Tiron-Tudor and Deliu (2022) formalize a Human-in-the-Loop framework positioning professionals as ethical governors of algorithmic outputs. Collectively these support an emerging conception of distributed ethical agency in which responsibility is shared among humans, organizations, and technologies — and explain why accountability, bias, and Trust concerns persist despite proliferating ethical guidelines.

Trust and relational ethics

Trust functions as a cross-sectoral ethical concern linking education, healthcare, agriculture, and business, with strong relational ties to decision-making and human–AI interaction. Importantly, it is not treated as a purely technical outcome of explainable algorithms: Maclure (2021) critiques reductive notions of explainability, arguing ethical legitimacy depends on institutional justification and public reason rather than computational transparency alone. This aligns with relational and affective analyses of AI in care contexts, and clarifies why qualitative, interpretive, and phenomenological approaches increasingly intersect with AI ethics research.

Education as a site of ethical contestation

Although healthcare dominates citation metrics, education emerges as a conceptually important context. Educational settings function as arenas where broader societal questions about intelligence, human Learner Agency, learning, and inclusion intersect with everyday professional decision-making. Baker et al. (2023) situate AI in education within philosophical debates about the "figure of the human," foregrounding questions of knowledge, authority, and inclusion. Teacher professional judgment is represented not as a technical skill to be replaced or standardized, but as an ethical and interpretive capacity central to responsible AI use — a view aligned with philosophical perspectives and the cultivation of professional judgment grounded in educational values.

Relevance to the knowledge base

This bibliometric paper provides field-level empirical grounding for the knowledge base's core themes around Ethics, AI in Education, and human-centered AI integration. It documents a documented disciplinary and conceptual shift toward practice-based, human-in-the-loop, and collaborative framings of AI ethics — reinforcing arguments that professionals should act as ethical agents rather than implementers of predefined frameworks. Its emphasis on Trust, interpretive practice, and professional judgment supports the knowledge base's treatment of AI not merely as a technical tool but as an ethically laden, context-sensitive technology. It also offers a methodological exemplar (bibliometric mapping) for tracing how the field's intellectual structure is evolving.

What this means for practice

• Researchers. Study AI ethics as situated professional judgment rather than compliance, and pair trend mapping with qualitative and interpretive inquiry into how educators actually negotiate ethical tensions.
• Researchers. Read citation counts as intellectual influence only; the authors state that highly cited work is not thereby practically or normatively more significant.
• Policymakers. Ground explainability requirements in institutional justification and public reason, not computational transparency alone — the chapter's reading of Maclure (2021) is that ethical legitimacy depends on the former.
• Policymakers. Resource teacher professional judgment as an ethical and interpretive capacity rather than standardizing it as a technical skill, which is how the educational literature in this corpus treats it.
• Policymakers. Watch that education's conceptual weight is not confused with its citation weight: healthcare dominates citation impact while education carries the interpretive debates.

Limitations

• Every one of the 282 records comes from a single database, Scopus, with the search executed on 31 December 2025, so coverage depends entirely on that index's inclusion criteria.
• The design is descriptive–evaluative bibliometrics: the authors state it does not synthesize empirical findings or assess study quality, and it supports no causal inference.
• Total citations and citations per year measure intellectual influence only; the authors explicitly state this limitation is inherent in citation-based approaches and qualifies how prominence of themes should be read.
• Screening excluded work that treated AI ethics solely at policy or regulatory level or without professional-practice reference, an eligibility judgment that shapes what the corpus can show.

Citation

Mazlan, C. A. N., Othman, M. A., Md Noor, A. R., Jamnongsarn, S., & Hidayatullah, R. (2026). From abstract ethics to situated practice: A bibliometric analysis of AI ethics and professional judgement. Educational Technology Research and Development.