dear Fathima Musfira Ameer

In my opinion, researchers and educators are addressing these challenges through multidisciplinary approaches that combine technical measures and instructional strategies. From a technical perspective, many support privacy-aware design frameworks (e.g. GDPR compliance, federated learning) and bias-reduction techniques (e.g. fairness-aware algorithms, diverse audits of instructional data). From an educational perspective, transparency is a priority – explaining AI decision-making to learners – while maintaining human oversight in the loop for critical assessments. Frameworks such as the EU Ethical Guidelines for Trustworthy AI or the UNESCO Recommendations on Artificial Intelligence in Education emphasize accountability, inclusion and learner agency. Practical practices include co-designing systems with educators, regular algorithmic audits and strengthening digital literacy to empower learners. Hybrid models that combine AI efficiency with teacher expertise are growing in popularity, ensuring that tools augment rather than replace human judgment. Ongoing collaboration between ethicists, engineers and educators remains key to balancing innovation with ethical rigor.

Debe orientarse hacia la promoción del desarrollo humano, el respeto por la privacidad, la equidad y la transparencia, guiado por principios pedagógicos y no solo tecnológicos.

Thank you, @Pedro Antonio Saltos García for giving your valuable perspective. I strongly agree that promoting human development, privacy, equity, and transparency guided by pedagogical principles is essential. Please kindly share any practical approaches or proven methods from research that help achieve these goals effectively in AI-powered learning environments. I’d love to learn more from your experience or knowledge.

Hamidreza Zahedi Thank you for your detailed perspective! I’m curious, from your experience, which specific practices or tools have proven most effective in maintaining transparency and human oversight in AI-driven assessments? I’m particularly interested in exploring how these multidisciplinary strategies can be effectively adapted within STEM education, especially in contexts with varying levels of digital literacy and infrastructure support.

In My Opinin,To maintain transparency and human oversight in AI-driven assessments, tools like explainable AI (XAI) frameworks—such as LIME or SHAP—help demystify decision-making processes, while audit logs and version control systems (e.g., Git) ensure traceability. Pairing these with human-in-the-loop (HITL) practices, where educators review AI-generated feedback or grades, fosters accountability. In STEM education, adapting these strategies requires scaffolding—for instance, using visual dashboards to interpret AI outputs for less digitally literate stakeholders or offline-capable tools (like lightweight LMS platforms) in low-infrastructure settings. Collaborative teacher-AI co-design workshops can also bridge gaps, ensuring tools align with local pedagogical goals while building trust through participatory transparency. The key is balancing technological rigor with culturally responsive training, ensuring both educators and students understand—and can challenge—AI’s role in their learning journey. :)

Interesting question and replies.

Since the focus is on ethics, the approach should have common practices with ethics in ICT use generally.

In my classes I tend to teach similar topics by engaging students in discourse on actual challenges or topic from actual incidents.

Konstantinos Karampelas Thanks for your thoughts. I recently did a study comparing AI tools, YouTube-based learning, and traditional methods for teaching software engineering concepts, especially engineering ethics. I have used TAM2 to see how students responded to each. Interestingly, AI tools were quite well accepted for their interactive nature, but there were still concerns around trust and over-reliance.

What worked best was blending AI with traditional teaching. This seemed to give the benefits of tech without losing the structure and guidance that students still value.

So yes, I totally agree that bringing real-world cases into the discussion helps a lot too. That said, my real interest lies in how ethical considerations can actually be applied in these settings. I’m especially looking into how we can develop a practical, adaptable framework that fits flipped classrooms, something that doesn’t just tick ethical boxes but truly supports responsible learning. Still trying to find the sweet spot.

By making AI models to be transparent in their operations and the data they use. This would help teachers understand how AI is making decisions and help students understand how their data is being used.

Secondly, there should be a clear policies and procedures that protect student data and prevent algorithmic bias.