Introduction

‍

In November of 2022, Open AI released ChatGPT — an open-access natural language processing tool that can write essays, draft content, conduct background research, generate computer code, and more. Its launch led to widespread concern throughout the higher education community that students would use this tool to “cheat” on coursework.

‍

At WGU Labs, where we are simultaneously technology enthusiasts and skeptics, we are excited to imagine what broad scale adoption of these new tools could do for learning. Though we are less concerned about students using the technology to cheat, we are very concerned about the implications for equity in higher education. Regardless of the current ethical uproar among faculty and administrators, students will take advantage of new technologies. Like nearly all new technologies we have seen before, the adoption of these AI tools will likely be faster and greater in more privileged populations, leaving underserved communities behind. With this in mind, we asked 3,143 students across nine institutions about their awareness and usage of ChatGPT in April of 2023 as part of our upcoming CIN Student EdTech Survey. Here are some more details of what we learned:

‍

• Key Takeaway 1: Awareness of ChatGPT is low, particularly among first-generation community college students.
• Key Takeaway 2:  Awareness of ChatGPT eliminates differences in utilization among first-generation students enrolled at four-year and primarily online institutions, but not community colleges.
• Key Takeaway 3: Few students represented ChatGPT output as their own work, but ethical perceptions of the practice depend on students’ prior usage.

‍

Methodology

In April of 2023, the CIN research team emailed surveys to more than 30,000 students across nine CIN member institutions. These post-secondary institutions included community colleges, private and public four-year institutions, and primarily online, not-for-profit colleges. The survey contained a variety of questions about students’ experiences with educational technologies, including five questions about their perceptions of and experiences with ChatGPT. Our final sample consisted of 3,143 students. Sixty-eight percent of respondents were enrolled at a four-year institution, 18.1% at a community college, and 13.5% at a primarily online college. Forty-eight percent reported that one or both of their parents/guardians graduated from a four-year university/institution, 10.4% reported that one or both of their parents/guardians attended (but did not graduate from) a four-year university/institution, and 39% reported that neither of their parents/guardians attended a four-year university/institution.

‍

Key Takeaways

‍

Key Takeaway 1: Awareness of ChatGPT is low, particularly among first-generation community college students

‍

We first asked students whether they knew about ChatGPT or other generative AI tools, and whether they had ever used these tools to help with their coursework. To examine potential inequities in awareness and utilization rates, we examined each of these questions by key student demographics, including parental education, institution type, race, and age. Although we did not find evidence of differences across race or age, we found important differences by parental education and institution type.

‍

Differences by Parental Education.

‍

Only about a third of students whose parents had never attended a four-year institution (i.e., first-generation college students) knew about ChatGPT. By comparison, roughly half of students with a parent who attended or graduated from college (i.e., continuing generation student) had heard about chat GPT.

‍

‍

Reported usage of ChatGPT was low across all groups of students, but particularly among first-generation students. About 7% of students whose parents had not attended a four-year institution reported using ChatGPT to help with coursework, while roughly 11% of students whose parents attended or graduated from a four-year institution did so.

‍

‍

Differences by generation status and institution type.

‍

The survey results also reveal that first-generation students’ awareness and usage of Chat GPT differed based on the type of college institution they attended. Results showed that first-generation community college students were less likely than first-gen students at four-year or primarily online institutions to report knowing about ChatGPT and other generative AI tools. Only 26% of first-gen students enrolled at a community college knew about Chat GPT, whereas 35% of those enrolled at a four year institution and 41% of those enrolled at a primarily online college did.

‍

‍

We found a similar pattern for usage. Only about 3% of first-generation students enrolled at a community college reported that they had used ChatGPT to help with their coursework. About 8% of first-generation students enrolled at four-year or primarily online colleges had used ChatGPT to help with their coursework.

‍

Interestingly, we did not see meaningful differences in awareness or usage by institution type for continuing generation students.

‍

‍

Key Takeaway 2: Awareness of ChatGPT eliminates differences in utilization among first-generation students enrolled at four-year and primarily online institutions, but not community colleges.

‍

When we asked students who were aware of ChatGPT whether they had used this tool for their coursework, we found no difference in usage between first-generation and continuing-generation students. This finding suggests that investing in education for first-gen students around new technologies may help foster more equitable usage.

‍

‍

However, we continued to see low usage rates among first-generation community college students, even when we restricted our analyses to those who were aware of ChatGPT. Among first-gen students aware of ChatGPT, around 20% of those enrolled at a four-year or primarily online institution had used ChatGPT, whereas only 10% of those enrolled at community colleges had. This suggests that additional work is needed to examine strategies to close usage gaps for first-generation students enrolled in community colleges.

‍

Key Takeaway 3: Few students represented ChatGPT output as their own work, but ethical perceptions of the practice depend on students’ prior usage.

‍

We next asked students whether they had ever turned in ChatGPT output as their own work for a homework assignment, course paper/essay, or exam. Our data showed that only about 3% of students in our overall sample had ever used ChatGPT to turn in ChatGPT output as their own work.

‍

‍

To further explore student usage, we also asked ChatGPT users in our sample about the specific ways they have previously used the tool. Among ChatGPT users, most reported using it in ways that enhance student learning, rather than using it to cheat on their coursework. The top three uses were to simplify complex topics, to brainstorm creative ideas, and to conduct research. About 31% of ChatGPT users in our sample reported using the tool to write homework responses or discussion points, and 23% reported using it to write papers and exams.

‍

‍

To assess students’ perceptions of the ethics of ChatGPT, we asked respondents the extent to which they agreed that “It is unethical to use ChatGPT to write responses to homework assignments, course papers/essays, or exams.” Across all students in our sample, the majority (54%) viewed ChatGPT usage in this way as unethical.

‍

‍

However, we also found that ethical perceptions differed by usage. ChatGPT users in our sample were less likely than non-users to view the tool as unethical. It is possible that these differences could be due in part to the fact that ChatGPT users are more aware of the ways that ChatGPT can be used to enhance learning and are therefore less likely to associate the tool with “cheating.” Because student ChatGPT users are primarily using the tool to simplify complex topics, they likely view it more as a research aid than a means to cheat.

‍

‍

Summary and Conclusions

‍

Based on this early view, ChatGPT, like many tech tools before it, is quickly finding its place among the tools students turn to to support their learning. But that uptake is still uneven — and uneven in ways that could amplify long-standing inequality in education.

‍

Despite widespread panic in the higher education community, we also found that very few students in our sample were using the tool to “cheat.” Instead, students reported using the tool to conduct research, brainstorm ideas, and simplify complex topics. This further points to the importance of educating underserved learners about generative AI to ensure they have access to the same learning supports as other student populations.

‍

ChatGPT and generative AI are not going away. Instead of prohibiting these tools, educators must empower diverse groups of learners to use them effectively and in ways that enhance the learning experience. Otherwise, we risk once again leaving underserved communities behind and unprepared for the tech-enabled future.

Without intervention, the imbalanced adoption of these tools could widen existing equity gaps. First-generation college students, and first-generation community college students in particular, were less likely than their peers to know about and have used ChatGPT to help with coursework. If these tools prove, as many anticipate, to occupy a prominent role in higher education and beyond, addressing these inequities will be critical to ensure that underserved learners can benefit from them to the same extent as their peers. ChatGPT and generative AI are not going away. Instead of prohibiting these tools, educators must empower diverse groups of learners to use them effectively and in ways that enhance the learning experience. Otherwise, we risk once again leaving underserved communities behind and unprepared for the tech-enabled future.