Abstract
Generative AI is as transformative as it is hazardous, especially in the education sector. Many students offload tasks to it to save time on both learning the process and completing the tasks themselves. Some use it to write emails and essays. Other students use it to summarize readings, explain topics, or check their work. All of these uses come with costs that students quickly realize. Often, when students offload their work to AI, they also offload the thinking process, which undermines their learning and critical thinking. In addition, AI is often incorrect or confidently states an incorrect answer when it really does not have the answer.
If Generative AI is as important as many are claiming, it becomes imperative to provide it as a tool to students in a way that avoids these costs. My technical capstone is one implementation that could do so by enhancing the ability of teaching assistants to aid students. My STS research paper explores how students are reacting to the costs and benefits of AI and argues that their behavior demonstrates a lack of regulation surrounding this technology.
The Teaching Assistant’s Teaching Assistant is an AI-supported office hours system that is designed to help TAs provide better and more efficient aid to students. Many computer science courses at UVA and other large universities can have a high demand for office hours. TAs can get overwhelmed trying to help all of the students, and may not have time to properly help each student. Additionally, a study showed that TAs used worse pedagogical techniques when they had to spend more time diagnosing the issue. The system is intended to address these issues.
In this system, students enter an online queue to join office hours, which is the custom for computer science classes at UVA. Students are asked to input their question or problem and any relevant context. Students in a mathematics course have used this context field to enter the problem, while students in computer science courses have used it to enter their erroneous code. When the student submits their input, they are entered into the queue, and their input is sent to an AI that generates a summary of the issue and possible misconceptions to the TA. Whenever a TA is done with one office hour session, they look at the online queue to see who is waiting, and on this screen they can read an AI summary of the issues the student is facing and thus prepare to help them. Data was collected from 658 office hours sessions. The primary findings were that students did not tend to enter enough information for the AI to determine an issue, but that it could be a helpful tool when they did.
My STS research consisted of fourteen student interviews. The interviews contained questions about students’ concerns with AI, their attitudes toward it, and the ways in which they use and choose not to use it. All students had concerns about AI, and all but two altered the ways in which they use AI as a result of their concern. Students mainly cited the degradation of their learning, the environmental impact, and misinformation as concerns. Many of them made comments about feeling hypocritical or guilty for using it.
These consequences observed in students can be connected to Ulrich Beck’s theory of Individualization. In an individualized society (as he claims developed western society is increasingly an example of), dissolution of social groups and class leaves responsibilities that were once faced by groups to be faced by individuals instead. Students, and AI users as a whole, have no social role that determines the ways in which they may use AI; this is a decision that must be made at the individual level in the absence of institutional regulation. However, many of the issues that stem from the development of AI cannot be solved by individual users, and thus lack a solution. The tendency of students to limit their own usage of the technology or feel bad for using it suggests a need for regulation at a higher level, perhaps in the laws of the federal or state governments, or in the rules of schools and universities.
