Effect of Operant Conditioning on Classroom Text-Generative Artificial Intelligence Attitudes

Peacock Scholarship

Public Deposited

Abstract

In this study, thirty-four participants took part in an experiment that measured the effects of operant conditioning on attitudes towards text-generative artificial intelligence, as well as usage of the same kinds of tools. Participants were asked to take a survey with written assignments asking about artificial intelligence's role in their education, as well as their hobbies. After the first written assignment, they were presented with an article praising artificial intelligence alongside feedback on the assignment that either was positive, negative, or neutral on it. Before and after the manipulation, an eighteen-question, seven-point scale was given to measure their attitudes towards artificial intelligence, and their use of artifiical intelligence in the second assignment was measured. It was hypothesized that both artificial intelligence use and attitudes would increase in the positive condition and decrease in the negative condition. Neither hypothesis was fully supported by the data, however, as both positive and negative conditions saw attitudes increase slightly and artificial intelligence use evenly split among participants. This is likely due to a lower-than-anticipated samle size and design flaws in the article's implementation that may have pushed all participants' attitudes in a similar direction.

Last modified