Initial tests of our proposed model with new GenAI motivation survey questions
For more information, see Trzesniewski, Gripshover, & Master (2026).
What happens shapes how I approach the next task
1How often isn't how
Students who use generative AI (GenAI) for school more often think with it more and have it do the work more. How often they use it isn't clearly linked to how often they catch its errors.
More frequent use → thinking with GenAI
More frequent use → having GenAI write or produce the work
More frequent use → catching GenAI making an error
The survey questions
- “In the past month, about how often have you used an AI chatbot (e.g., ChatGPT) specifically for school?” College 2
- “In the past month, how often have you… used AI to ask follow-up questions, push back on my ideas, or check my own reasoning.” College 2
- “In the past month, when using an AI chatbot for school, how often have you… caught the AI chatbot making an error.” College 2
- “In the past month, how often have you… had AI produce most or all of the work because you didn't feel like doing it.” College 2, one of three reasons
2Who can get good with GenAI?
Students who believe GenAI skills can't improve much more often use it just to finish fast, without revising what they ask or checking the response.
Believing some people could never become as effective with GenAI → finishing fast
Believing one's own GenAI ability will change very little → finishing fast
The survey questions
- “Even if they try, some people could never become as effective at using AI chatbots (e.g., ChatGPT) to improve their skills and knowledge as their peers.” agree or disagree
- “My ability to use AI chatbots (e.g., ChatGPT) to improve my skills and knowledge will change very little over time.” agree or disagree
- “When I use AI for school, I mostly want something fast that helps me finish. I usually do not spend much time revising what I ask or checking the response.” how well this describes me
3What instructors signal
Students who saw an instructor as believing only some students are smart more often wondered whether they belonged in that class.
Instructor seems to think only some are smart → doubt about belonging
The same, comparing the two classes each student rated
The survey questions
- “My instructor believes that some students are smart and others are not.” College 2, for each of two classes
- “How often do you wonder: “Maybe I don't belong in this class?”” College 2, for each of two classes
4Is it safe to use GenAI?
Students who had been accused of using GenAI when they hadn't were more often stressed about being accused. Students who worried about being accused more often avoided GenAI for fear of trouble.
Accused of using GenAI when they hadn't → stressed about being accused
Worry about being accused → avoided GenAI for fear of trouble
The survey questions
- “In the past month, how many times have you had each of these experiences in school? I was accused of using AI when I did not use it / I felt stressed or anxious about being accused of using AI / I avoided using AI to help me learn because I worried I'd get in trouble” high school and College 2
- “I worried about getting in trouble for using AI, even when I did not use it. / I avoided using AI because I feared I would get in trouble.” College 3, for each of two classes
Each dot is one sample: a national online panel of U.S. high school students (787) and undergraduates in three college samples (232, 293 and 163). The line through a dot is its 95% interval; a hollow dot means no clear link. Estimates are standardized and take into account background characteristics such as gender, race and ethnicity, first-generation status and year in school, and, in two college samples, how often students use GenAI (except where that is the starting point). The finishing-fast links also take into account how much students say they use GenAI to think. Arrows show the model's theorized order; the surveys were taken at one point in time, so they show what goes together, not what causes what.
Trzesniewski, K., Gripshover, S., & Master, A. (2026). Developing effective GenAI users: The missing motivational system and why it matters for equity. PsyArXiv. https://doi.org/10.31234/osf.io/uxh89_v1