Vol. 22:1 (2021) ► pp.55–80
What’s to bullying a bot?
Correlates between chatbot humanlikeness and abuse
In human-chatbot interaction, users casually and regularly offend and abuse the chatbot they are interacting with. The current paper explores the relationship between chatbot humanlikeness on the one hand and sexual advances and verbal aggression by the user on the other hand. 283 conversations between the Cleverbot chatbot and its users were harvested and analysed. Our results showed higher counts of user verbal aggression and sexual comments towards Cleverbot when Cleverbot appeared more humanlike in its behaviour. Caution is warranted with the interpretation of the results however as no experimental manipulation was conducted and causality can thus not be inferred. Nonetheless, the findings are relevant for both the research on the abuse of conversational agents, and the development of efficient approaches to discourage or prevent verbal aggression by chatbot users.
Article outline
- 1.Introduction
- 1.1Literature
- 1.1.1Incidence: Nonhuman conversation agents suffer more abuse than humans
- 1.1.2Ethical implications
- 1.1.3Need for a theoretical framework of chatbot abuse
- 1.2Current study
- 1.1Literature
- 2.Method
- 2.1Procedure
- 2.2Dataset
- 2.2.1Data collection
- 2.3Coding procedure
- 2.4Variables
- 2.4.1Coded variables
- Verbal aggression (Offence)
- Sexual references (SexRef)
- Nonsensical reply by Cleverbot (Nonsense)
- Insult by Cleverbot (Insult)
- Claiming to be (not) human (Humanity/Non-human user)
- Gender
- 2.4.2Other variables
- Turn count
- Approximate Loebner test (Loebner)
- 2.4.1Coded variables
- 3.Results
- 3.1Preliminary analyses
- 3.1.1Conversation descriptives
- 3.1.2Agreement
- 3.2Main analyses
- 3.2.1Verbal aggression
- 3.2.2Sexual comments
- 3.1Preliminary analyses
- 4.Discussion
- 4.1Limitations
- 4.2Conclusion
- Acknowledgements
- Notes
-
References