OpenAI recently launched a new large language model called ChatGPT that can generate original text—part of a brand of artificial intelligence (AI) known as “generative AI”—unleashing a barrage of panic, hype, and excitement. Such a mixed reception is par for the course with new AI tools, often met with hysteria about an impending dystopia and fears of job loss due to automation. As the most advanced chatbot to date, ChatGPT has led to catastrophizing about the potential for large language models to destroy higher education as we know it. Critics have decried the technology, arguing that it will make it possible for students to easily cheat, with little hope of detection, and will therefore undermine higher education. While it’s true that generative AI will likely lead to a reevaluation of teaching and assessment in higher education, changing pedagogical approaches can be a positive development. With ChatGPT and the development of other AI tools, educators will instead have to adapt and innovate to meet this new challenge. Rather than give into the panic about AI, higher education institutions should view these new models as an opportunity to rejuvenate the classroom learning experience.
In higher education, the college essay has largely been the center of humanities and social science curriculums. Many critics have specifically cited the potential of large language models like ChatGPT to destroy essay writing, as the model can write an essay for students based on a short prompt input. That potential has caused a plagiarism panic, as some educators worry that writing papers or answering homework questions by AI will become undetectable.
But this panic is misguided, as similar fears about technology have been in the past. In 2008, author Nicholas Carr argued that search engines make us “stupid” by diminishing our capacity as a species to concentrate. He was wrong—information technology innovation actually enhances our knowledge. Yet this type of neo-Luddite thinking is pervasive. Educators worried about the potential for cheating in the classroom with things like high-tech calculators, Internet search engines, and even spell-check. Instead of caving to cheating in these instances, schools had to adapt their approaches, such as limiting the use of calculators on tests and using plagiarism checkers like TurnItIn. ChatGPT and large language models should spur similar adaptations. New tools will emerge to detect this type of AI writing, which largely remains formulaic. Likewise, professors should adapt their essay prompts to focus on analytical thinking instead of regurgitating facts. They might also change the distribution of assessment types in a course to focus on proctored exams and fewer take-home tests.