Ganesh Mani

Distinguished Service Professor of Innovation Practice
Tepper School of Business
Spring 2025

46-992: MSBA Experiential Learning

Research Question(s): 

To what extent does students’ interaction with a consensus-building genAI tool (Thinkscape):

1. impact students’ self-reported ability to efficiently make group decisions?
   1. After completing both trials
   2. Immediately upon completion of each trial
2. impact students’ perceptions that their voices were heard during group deliberation?
3. impact students’ perceptions of the quality of groups’ decisions?

Teaching Intervention with Generative AI (genAI):

As part of their Experiential Learning Capstone course, Mani had his students complete collaborative resource allocation tasks that challenged them to arrive at a group consensus in an efficient manner. Students worked in small groups using a consensus-building genAI tool (Thinkscape) and had 30 minutes to arrive at a group consensus regarding how to allocate a financial portfolio based on the profile of a potential client.

Mani taught the course in the Spring 2025 semester. First, students used a whiteboard to make small-group decisions for a resource allocation task. A week later, students used the genAI tool (Thinkscape) to complete a similar resource allocation task for a different profile. This design allowed all students to experience both conditions and serve as their own point of comparison. Students completed a brief follow-up survey to gauge their perceptions after each activity. At the conclusion of both tasks, students were also asked to compare the two modalities in terms of perceived efficiency.

Sample size: 30 students completed the control task, followed by the treatment task

Data Sources:

1. Student survey data collected immediately following each activity and at the conclusion of both trials.

1. RQ1a: When surveyed at the conclusion of both trials, students were asked to choose which modality more efficiently helped their groups reach a conclusion. Students indicated that the Thinkscape trial (67%) more efficiently helped them reach a conclusion, compared to the whiteboard trial (13%).

   
   
   

   Figure 1. At the conclusion of both trials, 67% of the students thought Thinkscape more efficiently helped them reach a conclusion compared to using a whiteboard (13%). A Chi-Square Goodness of Fit test revealed these differences to be significant, χ²(2) = 15.20, p < .001.

2. RQ1b, 2 &3: When surveyed with Likert-type items immediately following each trial, students did not report a significant difference between the conditions for the perceived efficiency of the process, perceptions of their voices being heard, or the perceived quality of the decisions their group made.
   
   

   Figure 2. There was no significant difference between students’ ratings of efficiency t (29) = 0.96, p = .34, feelings of their voice being heard t (29) = 1.31, p = .20, or perceived quality of the groups’ decisions t (29) = -1.20, p = .24 when completing the allocation task with a whiteboard compared to using the genAI tool.

Eberly Center’s Takeaways:

1. RQ1, RQ2, & RQ3: At the conclusion of both trials, students were asked to choose between the two modalities in terms of efficiency, and a majority of the students chose the Thinkscape genAI tool over the whiteboard. When rating each trial independently, however, students’ perceptions regarding group process and decision making did not differ between the use of the genAI tool and a traditional whiteboard. All of these students experienced the whiteboard trial first, followed by the genAI trial, so we cannot say the extent to which ordering effects impacted any of the results. There were plans to include a second section that completed the trials in a counterbalanced order (genAI first, followed by whiteboard), but low attendance at one of the trials prevented that section from being included in this analysis.