What Robot Jokes Are Funniest? Study Finds Aggressive and Affiliative Humor Win
Would you laugh if a robot told you an AI-generated joke? A study published in June ran real-robot experiments in a university classroom, evaluating how different humor types and joke topics affect perceived funniness and acceptability. The results show that humor type significantly drives funniness, with aggressive and affiliative humor rated highest.
- The study had a robot deliver AI-generated jokes in a university classroom and evaluated perceptions of different humor types
- Humor type significantly influences funniness, with aggressive and affiliative humor rated higher
- Joke content (person-related vs. political) shapes acceptability
- The experiment also examined participants’ language preference in a bilingual setting
Experiment Design
The study used a mixed factorial design in which participants evaluated AI-generated jokes delivered by a robot in a classroom. Variables included four humor types, affiliative, self-enhancing, aggressive and self-defeating, plus two joke topics: person-related and political. Participants were also asked which language they preferred.
Results: Which Jokes Land Best?
Results show humor type significantly influences funniness, with aggressive and affiliative humor rated higher; joke content shapes acceptability, with political jokes being more sensitive. The finding has practical implications for human-robot interaction: if an AI robot wants to tell jokes in social settings, choosing the right humor style matters more than the content itself.
Implications for AI and Human-Robot Interaction
This is an exploratory study in human-robot interaction (HRI), showing that LLM humor generation has matured enough to be tested on real robots. If future AI assistants and service robots add humor, this kind of research helps design jokes with a sense of proportion, funny without offending. For AI developers, humor is not just an entertainment feature; it is part of building trust between humans and machines.
