First-Year Engineering Students' Expertise and Trust in GenAl

This full-length research study investigated first-year engineering students “trust in generative artificial intelligence (GenAl) before and after course instruction. Pre-and post-surveys were conducted with questions on students” experience with GenAl tools as well as trust in GenAl. The trust questions had students evaluate the likelihood of GenAl generating a correct response to various prompts such as “explain the unit circle” (correct response likely) to “solving this system of equations …” (correct response unlikely for tools available in Fall 2023). Within-subjects analyses indicated that lessons in the course significantly increased trust in ChatGPT for correct-response-likely items. In addition, the lessons significantly decreased students “trust in ChatGPT for correct-response-unlikely items. There were no significant interactions between prior experience level and change in trust. These results show that guided exposure to GenAl helped first-year engineering students begin to understand capabilities and limitations of GenAl. These results are promising because student trust in GenAl output will directly impact decisions to engage with it for different tasks. More work is needed to optimize instruction and understand students” use of the tool beyond the classroom integration, including their full ethical decision-making process, but the malleability of trust at this level is an indication that engineering educators can impact student perspectives of GenAl.