4 Comments
User's avatar
Lewis Kang'ethe Ngugi's avatar

This is a brilliant analysis! I recall these principles being discussed in Erika Hall's book: Conversational Design under the 'Principles of Conversational Design' section. What I'm particularly curious about is finding the most streamlined approach to gathering user feedback on chatbots based on these maxims. As designers, we aim to minimize user friction while still collecting meaningful data, what do people consider the leanest possible way to have users rate chatbot responses against each of Grice's maxims?

https://www.muledesign.com/blog/conversational-design-for-you

Expand full comment
Cristian Pineda's avatar

Thanks for sharing this, Lewis! It's an interesting challenge that I believe requires a multimodal approach, tailored to different stages of the user journey throughout the product lifecycle.

A thumbs down feedback experience is one that comes to mind, I see a strong opportunity to enhance it. We could include quick actions that allow users to specify which of Grice's Maxims were affected—perhaps framed in a more user-friendly tone. This would provide richer insights while maintaining an accessible interaction for the user.

Looking ahead, I also envision the possibility of leveraging a specialized model to automatically evaluate instances where users give a thumbs down, by analyzing the prompt and response directly to assess potential issues against our Maxims.

I hope these added thoughts spark some ideas! And thank you for mentioning Erika Hall's book; I'll make sure it's added to our community library of shared resources.

Expand full comment
Nimisha Patil's avatar

Such a valuable & actionable framework! Thanks for the post Cristian.

Expand full comment
Cristian Pineda's avatar

Thank you Nimisha! Glad you found it useful :)

Expand full comment