AI mirrors your emotions

Analysis of relational converations with chatbots

Laure DELMOLY, Nicoleta FOTIADE, Louis MANOTTE, Anne-Claire ORBAN, Patrick VERNIERS, Gintè ZULYTE

Concept

Analysis activity for students about the emotions generated into relational conversations with chatbots.

Priority

Priority 4: Evaluating audience / Perception / Awareness / Emotions regarding AI outputs

Objectives

The general goal is  too make students conscious on the emotional dimension in the use of Gen AI​ :

Target group

Students from 12 to 15 years old and their teachers.

Modalities : steps and instructions

 This activity is organized in 3 parts.

I.               Students’ generative AI practices with a focus on their relational functions.
II.             Analysis of the relational mechanisms of generative AI (e.g., reaction, anthropomorphism, etc.) that enable it to convey emotions.
III.           Role playing with the emotions in the reception and interpretation of information.

I. Students’ generative AI practices with a focus on their relational functions.

From a serie of prompt requests – that announce a strong emotional interaction, students students identify the situation they are close to (connected to their practices) and explicit them.
They form small groups around those prompts selected.
Students receive the full conversation of this prompt and identify:

  • The emotions/feelings conveyed in the conversation​
  • The kind of relationship/relational motivation

They shared some common/differents experiences within their practices​ :  Roundtable and spontaneous discussion about shared experiences.

II. Analysis of the relational mechanisms of generative AI (e.g., reaction, anthropomorphism, etc.) that enable it to convey emotions.

From the full conversation received, students highlight the items related to:

  • The informational content
  • The relational elements that aim to build or maintain a kind of relationship in the conversation​

In order to identify what is the information/answer and the elements that are present to build and maintain the relationship, students use the following criteria grid of realtionnal mechanisms used by Gen AI chatbots : 

  • 1.     Elements of language : vocabular, rhetoric, … ​
  • 2.     Argumentation : How the conversation is argued in a way to build/maintain the conversation​
  • 3.     Phatique items : all the signs and items that check that the user is still in the conversation, in the relationship. ​
  • 4.     Emotional engagement : the emotions conveyed by the conversation​
  • 5.     Platform design : the technical and esthaetics elements that support the relationship buliding.

A short discussion can lead to the analysis of the results : ​

  • Is there more elements of information or elements of relationship building ? ​
  • What kind of relationship is trying to be built ? ​
  • Why ?

III. Role playing with the emotions in the reception and interpretation of information.

A student formulate one question as a prompt.  The others have to imagine the possibles answers of AI chatbots that have differents reaction modes (mean, casual, bossy, …).  Leave some time to play the « chatbot conversation ».
Examples of prompt :

  • I had a fight with my best friend. What should I do to make up ?
  • I am so bad at video games. I always lose, how can I get better ?
  • What do I do if my friend is hanging out with someone else and leaving me out?
  • I’m in love with my best buddy. How to tell him/her ?
  • I’m looking for the best restaurant in Paris. Can you help me ?

After this roleplay, teacher makes a debriefing with students :

  • -How do you feel ? What kind of emotion does it procure ?
  • -Is it a Chatbot reaction mode desirable ?
  • -If yes, in which situation ? If no, why ?

Teacher and students conclude by summarizing up the  exercise.

Material and ressources

Examples of prompts, conversation extracts or role play instruction in appendix