AI evidence extraction

Ethical challenges of artificial intelligence in education: A systematic literature review on bias, privacy, and academic integrity

Musah Ebikabowei · International Journal of Applied Resilience and Sustainability · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.70593/deepsci.0202054

Methodology & findings

Study design

Systematic literature review following PRISMA 2020 guidelines.

Main result

The review finds that "artificial intelligence in education has reached a radical stage where the technological innovation is increasing at a rate that is higher than the ethical governance, institutional readiness and policy formulation." The most pressing ethical issues identified are algorithmic bias, data privacy and surveillance, and academic integrity challenges, particularly with generative AI. The paper notes that "Diffusion and Generative Models, alongside Large Language Models, cluster conspicuously in the high-bias and high-privacy-risk quadrant, signalling their dual ethical burden in educational deployments" while "Federated Learning models concentrate in the lower-left quadrant, reflecting their architectural design advantages in mitigating both bias amplification and data exposure."

Research paradigm

Critical interpretivism with systematic evidence synthesis

Author conclusions

The authors conclude: "This is a structural literature review that indicates that artificial intelligence has entered education at a relatively quicker rate than ethical government, institutional preparedness, and regulatory control. In all the reviewed pieces of literature, the most prominent ethical issues connected to AI in education are bias, privacy, and academic integrity, especially when it comes to generative AI, large language models, learning analytics, automated assessment, and predictive educational technologies." They further state: "Finally, it is up to the educators, policymakers, technology designers, and institutions to design human-centered, reliable, and ethically responsible systems that can assist learning without violating the privacy, integrity, or inclusion."

Risk of bias

Publication bias: research concentrated in high-income nations and elite institutions; Selection bias: limited representation of K-12 and low-resource educational contexts; Database selection bias: search limited to four databases, may miss grey literature; Language bias: non-English publications explicitly excluded; Temporal bias: long-term effects of AI on student outcomes not captured; Geographic bias: majority of research from high-income countries and elite institutions; Language bias: search limited to English-language publications; Publication bias: only peer-reviewed sources included in final synthesis (non-peer-reviewed sources excluded); Scope bias: focus on three ethical dimensions (bias, privacy, academic integrity) may miss other relevant ethical concerns; Selection bias: exclusion of non-empirical opinion pieces and studies without substantive ethical analysis; Geographic bias: majority of reviewed literature from high-income countries and elite institutions; Publication bias: limited research on low-resource settings and K-12 education; Language bias: only English-language publications included; Database bias: search limited to four major databases which may exclude grey literature; Selection bias: exclusion of non-empirical opinion pieces may miss important conceptual contributions; Temporal bias: literature review window not explicitly specified for all searches

Limitations

The authors state: "Most of the existing research is focusing in universities and high-income nations and there is still limited research on low-resource learning environments, K-12 education and addressing culturally diverse conditions
Also, the long-term effects of AI on the well-being of students, educator autonomy, and student cognitive growth, as well as and educational disparity, are not sufficiently empirically evidenced
There is also a lack of unity in policy implementation and regulatory systems with a wide disparity between institutions and nations." Additionally, "Geographic gaps are also being identified in existing research since the bulk of the research has been conducted in the high-income countries and elite institutions, and little is known about the effects of the algorithmic bias in low-resource settings, rural schools, and educational systems in developing countries."

Open questions raised

Limited empirical evidence on long-term effects of AI on student well-being, educator autonomy, and student cognitive growth
Insufficient research on low-resource learning environments, K-12 education, and culturally diverse educational conditions
Lack of stakeholder engagement research, institutional readiness, and perspectives from students, teachers, and policymakers in various educational settings
Missing evidence on algorithmic bias effects in low-resource settings, rural schools, and developing countries
Need for interdisciplinary ethical governance frameworks and culturally encompassing AI systems
Requirement for more empirical investigation of underrepresented areas and evidence-based fairness, transparency, and accountability studies

Extracted from: pdfAgreement 82%

Explore related topics

Responsible AI Use Research Integrity Literature Review