Date
Publisher
arXiv
Group-based ethics training for Science, Technology, Engineering and Mathematics (STEM) students is a complex challenge, requiring substantial resources and expertise. While activity-based teaching methods, such as role-playing and discussions, are commonly employed to simulate real-world scenarios, current practices are often manual and lack integration with effective online platforms for supporting group-based ethical discussions. In this work, we propose Ethics Training Agents, a group discussion system that leverages multiple LLM participants embodying distinct ethical orientations, along with a moderator agent, to enable structured human-AI group ethical discussions for collaborative reflection. We conduct a user study with 45 undergraduate STEM students to evaluate the learning outcomes and user experience. The results show that our system supports engagement, coordination, and perspective-taking in group discussions and has a positive influence on ethical sensitivity. We also discuss practical design strategies for integrating multiple LLM agents into multi-human group settings to facilitate ethics training for STEM students.
Study at a glance
- Question: How do discussions with AI agents and human peers affect STEM students' ethical sensitivity and discussion dynamics?
- Setting and data: University classroom in Republic of Korea; 45 undergraduate STEM students across 15 groups; pre- and post-surveys, discussion logs, and focus group interviews.
- Method: Mixed methods user study using pre-post surveys, focus group interviews, and log analysis without a control group.
Why it matters
STEM curricula often leave students underprepared for technology's societal consequences, but existing ethics education lacks scalable group-based methods that provide consistent learning experiences without substantial instructor facilitation.
Key findings
- Undergraduate STEM students showed statistically significant gains in both ethical issue awareness and ethical consequence reasoning after participating in the AI-facilitated discussions.
- Participants rated human peers significantly higher than AI agents across general contribution, diversity of ideas, and influence on thinking.
- Participants directed 79.7% of their questions to AI agents rather than human peers, finding AI less emotionally burdensome to question despite valuing human responses more.
Summary drafted with AI from the full paper and may contain errors. See the full paper for details.
What is the application?
Who is the user?
Who age?
Study design
