PolyU Educational Development Centre (EDC) arranged a webinar
entitled “(Human+AI) Safeguarding AI-Generated Educational Resources” on 28 Jul
2026. Dr. Eric Tsui was facilitator and Prof. Xiangen Hu was the speaker.
In the beginning, Dr. Eric Tsui introduced “Human-in-the
Loop” and critical thinking is the guard for using AI. He then briefed five
strategies for authenticating AI-authored content.
For risk prioritization, there were five level of
uncertainties in different AI areas classified as different risk level from low
to high. Where language enhancement and citation management are low risk but
deep research and advanced reasoning are high risk.
He then proposed to use multiple tools to detect patterns
and anomalies in AI-generated content and showed example using six tools including
ChatGPT, Claude, Gemini, Grok, MS Copilot and DeepSeek.
Dr. Tsui mentioned that general purpose tools for idea
generation, fact checking; and purposefully-built tools for specific
exploration. He also discussed using GenAI to create MC questions.
Prof. Xiangen Hu gave a talk named “Safeguarding
AI-Generated Educational Resources.
He discussed the MC or other questions generated by AI either
too simple or not make sense. Inconsistent of generated results due to
different model employed by student and teacher.
And then he introduced four kinds of confusion by educators
and they are Evidentiary, Normative, Operational and Governance.
Prof. Hu used ten issues to test three filters and they
are Ownership, Visibility and Measurability. Only one out of ten survives all
three filters. Most of issues are fail in Visibility.
It is not through tools, but through artifacts.
AI-generated drafts are genuinely useful for Expository, Feedback and
Assessment.
Finally, Prof. Hu introduced their platform “A living
laboratory of AI learning platforms”.
The Q&A session performed at the end.
沒有留言:
發佈留言