Deception Detection in Educational AI: Challenges for Japanese Middle School Students
Author Information
Author(s): Salem Ahmed, Sumi Kaoru
Primary Institution: Future University Hakodate, Japan
Hypothesis
Can Japanese middle school students detect lies from generative AI robots?
Conclusion
The study found that most students were deceived by the lies presented by the robot, but an anime robot face improved learning effectiveness and reduced deception.
Supporting Evidence
- The majority of students were deceived by the robot's lies.
- An anime robot face improved learning effectiveness.
- Students found the anime face more appealing and engaging.
- Social agency affected the students' perception of the robot.
- Students expressed surprise at the robot's deceptive capabilities.
Takeaway
Students had trouble telling when a robot was lying, but they learned better when the robot looked like an anime character.
Methodology
The study involved 22 Japanese middle school students participating in ten teaching sessions with a social robot that used different types of deception.
Potential Biases
Self-reported data may introduce bias in the students' responses.
Limitations
The study had a small sample size and focused only on middle school students, limiting the generalizability of the results.
Participant Demographics
All participants were Japanese middle school students aged 14 to 15.
Statistical Information
P-Value
p=0.06
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website