ChatGPT, OpenAI’s fast-growing language model that can write everything from essays to poems and even computer code, is roiling classrooms from middle school to graduate school, leading school districts across the country — including in Seattle, Los Angeles and New York City — to ban its use. But should they?
Good writing has no shortcut. It requires revising, reflecting and revising yet again. Every semicolon matters. Every comma contributes to the rhythm of a piece, to the way in which a writer conveys deeper meaning. Just watch the new documentary, “Turn Every Page,” where you’ll see Pulitzer Prize-winning author Robert Caro and his talented editor Robert Gottlieb reveal how human beings think, and even agonize, through the process of writing.
ChatGPT seemingly helps a writer leap to the finish line, avoiding the mental sweat involved in putting together words to bring ideas to life. But that very work — that journey — is necessary if writing is to transform thinking and lead to answers to humanity’s pressing problems.
When we asked ChatGPT about itself, it spat out an answer in a couple of seconds: “People may become too reliant on ChatGPT and other language models, and may stop putting in the effort to think critically and creatively.”
We work at universities, and have specialties that make us believe in the power of clear, accurate prose. We teach and research writing, technology, child development and media. AI language systems affect three crucial purposes of writing: to transmit information, to conduct transactions and to transform thinking.
We have seen that ChatGPT productively supports the more mundane uses of writing — transmitting existing information and conducting transactions, like drafting a legal contract, crafting a cover letter or resume, producing a simple business plan, or writing an email to dispute a credit card charge. It saves time, so long as humans revise and fact-check before anyone even considers distributing or publishing the prose.
Struggling to come up with a business proposal touting a 2-in-1 vacuum and skateboard? Just ask ChatGPT to create your pitch:
Introducing the “SkateVac” — a revolutionary vacuum cleaner that doubles as a skateboard, making cleaning your home fun, efficient and eco-friendly by using the power of your own movement.
As impressive as it is, ChatGPT hasn’t adapted to include a human’s sense of nuance. It doesn’t care about the semicolons, it doesn’t wrestle with word choice. It doesn’t care what you want to say.
ChatGPT is an optimal synthesizer. In a world flooded with information, it is almost impossible to assemble all the sources available into a coherent narrative. Right now, these bots are completely incapable at writing with nuance and perspective. Yet for a very rough first draft, this type of organizational tool that produces cohesive text, can prepare writers for a higher-level of engaged, transformative writing — so long as they add critical thinking and a creative touch that eludes artificial intelligence.
Should universities and workplaces halt its use? That’s futile. Let’s learn how to ask the right questions — then reflect, revise and revise. Let’s learn how to check our sources. Let’s use ChatGPT as a platform, a starting point and not a final destination. First draft, “yes.” Final draft, “no way.”
Joshua Wilson (email@example.com) is an associate professor of education at the University of Delaware’s College of Education and Human Development researching ways that technology can transform the teaching and learning of writing. Also contributing to this piece from the same college division are doctoral student Amanda Delgado, who is researching children’s engagement and learning from storybook reading and digital media; and Roberta Michnick Golinkoff, the Unidel H. Rodney Sharp Professor at the university, where she holds appointments in the School of Education, Department of Psychological and Brain Sciences, and Department of Linguistics and Cognitive Science. Golinkoff is also co-author of “Making Schools Work” (Teachers College Press, 2022) with Kathy Hirsh-Pasek, a professor of psychology at Temple University and senior fellow at the Brookings Institution who also contributed to this piece.