Table of Links
Abstract and 1. Introduction
- Methods
- Quantitative Results and Creativity Support Index
- Qualitative Results from Focus Group Discussions
- Discussion
- Mitigations and Conclusion and Acknowledgments
- Ethical Guidance References
A. Related Work on Computational Humour, AI and Comedy
B. Participant Questionaire
C. Focus
C FOCUS GROUP QUESTIONS
C.1 Qualitative questions about the specific writing task
• Did you find any of the generated outputs helpful? If so, could you recall one output that was usable and explain in what way it helped you write? • Could you recall one generated output that was not usable, and explain why?
• How do you think the generated output differed from what you would find in some of the resources you use often, e.g., Wikipedia, Google search, other artists’ material?
• How did the prompts we suggested differ from your custom prompts?
• Comment on the types of comedy that you managed to generated with the AI tools.
• Were any of the generated outputs that were presented offensive/inappropriate in some way? If so, what did you think of these?
• What made you decide to stop generating outputs?
• Did you have any concerns about ownership or agency when generating outputs?
C.2 General discussion points about AI tools for comedy
• What does your comedy writing process look like? How is this process of working with AI different from working alone or working with other comedians?
• What can you say about the stereotypes in the outputs generated by the AI writing tool? How do they relate to or differ from stereotypes in human written comedy?
• Is humor and comedy evolving over time? Should AI tools adapt to societal changes in a similar way?
• Does the computer have a “voice”? How would you compare it with the voice of a human comedian? What are the consequences for writing comedy about identity?
• What is your stance on moderation? Should the output of AI tools be allowed to be edgy? Can and should the output of AI tools be moderated?
• How does context shape the meaning of comedy? Who has the responsibility for the comedy material?
• When is it acceptable to use other comedians’, or in general, other artists’ work? What about the outputs of an AI writing tool?
• What importance do you attach to the text vs. to the delivery of that text? Could a human comedian make AI-generated content better?
• Can comedy and humour be quantified and measured?
Note that the leading questions in the focus group could be considered as biasing the respondents. We chose this formulation to be able to address the topics of interest listed in Section 1.2.
C.3 Debriefing questions
• Is there anything you’d like to share that I didn’t ask about?
• Is there anything you’d like to know or ask me?
:::info
Authors:
(1) Piotr W. Mirowski∗, Google DeepMind London, UK ([email protected]);
(2) Juliette Love∗, Google DeepMind London, UK ( [email protected]);
(3) Kory Mathewson, Google DeepMind Montréal, QC, Canada ([email protected]);
(4) Shakir Mohamed, Google DeepMind London, UK ([email protected]).
:::
:::info
This paper is available on arxiv under CC BY 4.0 license.
:::
n
