Primer: Artificial Intelligence
Inaugurated by the release of the OpenAI’s ChatGPT in November 2022, the recent ascendance of generative artificial intelligence (AI) has had a transformative effect on many facets of daily life. Perhaps nowhere has this transformation been more acutely felt than in education. Across campuses and classrooms around the globe, students and teachers alike have had to grapple with both the pendent dangers and the incredible potential of what is, by all accounts, the defining technological innovation of our time.
What separates the large language models of the present, such as ChatGPT, from the artificial intelligence of the past is their capacity to generate human-like language through training on vast amounts of text. These systems predict and produce words in sequence, enabling them to compose essays, summarize readings, solve problems, and even simulate conversation with remarkable fluency. In the sphere of education, their potential is both expansive and contested.
As a consequence of these new AI models’ ability to generate and modify academic writing, a great number of students have begun to appropriate AI as tools for coursework, writing essays, solving homework problems, and even composing discussion responses with little to no personal input, casting in a new form longstanding concerns of academic integrity and authorship. While schools insist, as a matter of pedagogical and ethical principle, that students must produce their own work, the rise of AI has made that mandate freshly complex. Just as cheating on an exam by using another student’s answers is punishable because it undermines the integrity of assessment, so too does the unacknowledged use of AI challenge the foundation of honest scholarship. Whereas plagiarism once involved the copying of another person’s work, AI enables a new form of misrepresentation: the production of text that is perhaps original in form but not in thought.
Proponents see in artificial intelligence another tool for the benefit of education, in the vein of the calculator or the internet. They argue that adaptive AI systems can provide a kind of individualized attention, tailored to the particular problems and questions of individual students, that most classrooms, particularly those in under-resourced schools, can hardly provide. AI, thus conceived, has a sort of incredible egalitarian function: an omnipresent teacher capable of assisting anyone, anywhere. It holds out the promise of democratizing education, much as the internet or Wikipedia have democratized access to information, an unprecedented expansion of learning’s reach, though not, perhaps, of its depth.
On the other hand, critics warn that easy access to AI fosters a dependency that risks the degradation of students’ critical faculties. Rather than grappling with questions themselves, students substitute genuine thought for the convenience of generated answers. The concern is not merely that they might let AI do their work but that they might let it do their thinking. Moreover, the critics also note that AI models carry their own biases, arising from both from the selection of information on which they are trained and from their particular constitution, which can be inimical to education in a way the varied and heterogeneous biases of human teachers are not.
Regardless of one’s general opinion, it is difficult to deny that artificial intelligence brings both benefit and harm to education. Any lucid appraisal of the effect of AI on education must therefore ask whether its promise of convenience and accessibility justifies the quiet diminution of the very habits of mind that education exists to cultivate.
Please come join us Monday night at 7pm in Scott Hall 201!
“ChatGPT” by Ged Carroll, CC BY 2.0