The students of Trinity College were given a new yet challenging task. In a recent January Term course titled "Writing and AI," visiting assistant professor Alex Helberg asked the students to write persuasive essays using generative AI chatbots.
The three-week J-Term course aimed to explore the capabilities of contemporary chatbots, such as OpenAI’s ChatGPT, Microsoft Copilot, and Google’s Bard, which have gained attention for their applications in various fields.
Helberg explained that large language models, like these chatbots, essentially refurbish existing text from the internet without understanding, needing students to strategically input prompts to achieve the desired output, also called prompt engineering.
“They don’t actually understand anything and can only produce variations of what humans have already produced,” said Helberg.
The course began with an overview of the history of AI, leading up to the latest AI programs. The assignment tasked students with crafting a "synthetic paper" advocating for a significant improvement to Trinity's curriculum or campus. The challenge was to ensure that all information in the paper was verifiably accurate, avoiding the generation of misinformation.
The resulting generative AI-written papers were accompanied by journals detailing the input prompts and generated output, along with reflective essays where students assessed the process and contemplated the potential use of AI in their own writing.
Helberg said that the course's purpose was to prompt philosophical considerations about the implications of AI for humanity and demystify misconceptions surrounding AI's capabilities. The class also delved into ethical and policy issues related to writing and AI, particularly in educational settings.
“The paper’s theme was to argue for a change to the Trinity curriculum or campus that would significantly improve the student experience, but every piece of information in the paper has to be verifiably accurate,” Helberg added. “The trick here is to strategically engineer input prompts so that your synthetic text technology is not generating misinformation.”
While acknowledging that synthetic text generators do not signal the end of human-centered writing, Helberg believes that these tools help in brainstorming and ideation in the writing process. Students expressed surprise at the limited intelligence of synthetic text generators, dispelling media hype.