Student and Faculty Perceptions of Artificial Intelligence in Student Writing
Autumn B. Hostetter, Natalie Call, Grace Frazier, Tristan James, Cassandra Linnertz, Elizabeth Nestle, Miaflora Tucci
Abstract
Artificial Intelligence (AI) can write poetry and news articles that go undetected by human readers. Can students use AI to write college assignments that go undetected by their professors? Past and current perceptions of AI in education have differed; while some people view AI as a tool, others view it as a threat to learning. We surveyed 83 students and 82 faculty, providing them with 4 writing samples, 3 generated by undergraduate students and 1 generated by the AI-chatbot, ChatGPT-3. We found that neither faculty nor students could detect AI-generated writing at above chance levels. Faculty and students had similar opinions on the ethicality of various uses of AI technology and how much these uses are likely to compromise learning. Faculty reported a high level of concern regarding the potential effects that AI could have on their pedagogical practices. Prior experience with ChatGPT-3 and analyzing the structure and organization of the response was found to improve detection ability in faculty, suggesting that increased exposure and domain-specific analysis may be beneficial in the detection of AI in students’ writing.