EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An ethical AI framework for STEM education

Meysam Abedi, Ismaila Temitayo Sanusi, Markku Tukiainen

American Journal of STEM Education · 05.07.2026 · 25 · 121-146

DOI: 10.32674/04054e51 ↗

Kurz gefasst

Fachartikel in der Zeitschrift American Journal of STEM Education, online erschienen am 05.07.2026. Verfasst von Meysam Abedi und 2 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study introduces a novel ethical AI framework for undergraduate STEM education that prioritizes privacy, transparency, and accountability. Employing a convergent parallel mixed-methods design, this study engaged 412 undergraduate STEM students, 15 faculty members, and 10 administrators across three universities over two semesters. Data collection integrated quantitative learning analytics with qualitative stakeholder interviews and focus groups to capture both measurable outcomes and lived experiences. The results demonstrate that responsible AI design significantly improves student engagement (35%), instructor acceptance (78%), and reduces performance gaps between student groups by 40%, all while maintaining the model's predictive accuracy at 89%. This research demonstrates that AI can be designed to be both technically robust and ethically committed.

Zitieren

APA

Abedi, M., Sanusi, I. T., & Tukiainen, M. (2026). An ethical AI framework for STEM education. American Journal of STEM Education, 25, 121-146. https://doi.org/10.32674/04054e51

Vancouver

Abedi M, Sanusi IT, Tukiainen M. An ethical AI framework for STEM education. American Journal of STEM Education. 2026;25:121-146. doi:10.32674/04054e51

BibTeX
@article{Abedi2026,
  author = {Abedi, Meysam and Sanusi, Ismaila Temitayo and Tukiainen, Markku},
  title = {An ethical AI framework for STEM education},
  journal = {American Journal of STEM Education},
  year = {2026},
  volume = {25},
  pages = {121--146},
  doi = {10.32674/04054e51}
}
Kostenlos starten →