The stigma attached to using LLMs is valuable. But it likely masks the extent to which younger Americans have come to rely on AI in all sorts of ways.
When asking young adults about their ChatGPT habits, you’re likely to hear vigorous assertions that they scrupulously avoid using the application — or, at least, that they don’t use it all that often. The stigma that has developed around this digital tool is valuable. That stigma, however, likely masks the extent to which younger Americans have come to rely on this and other large-language-model AIs to do their thinking for them.
Researchers at the University of Chicago recently attempted to quantify the extent to which young people rely on LLMs — a project that must clear away the “social desirability bias” against relying on ChatGPT to do the work of basic cognition. For example, the researchers found that about 60 percent of surveyed students admit to using AI tools themselves. And yet those same respondents believe that 90 percent of their peers use AI and LLMs. “The majority of students report using AI either just ‘a little’ or ‘not at all,’” the researchers observed. That stands “in contrast to others’ use, where the two largest categories are using AI ‘a moderate amount’ or ‘a lot.’”
We can safely deduce from this that the students who insist they never or rarely use AI, in defiance of the trends set by their peers, are understating their own reliance on LLMs. But what are they using them for?
In a useful exploration of the phenomenon, Pace University communications student Lucy Anderson assembled a collection of anecdotes with the aim of exploring the degree to which her peers have come to rely on ChatGPT and other similar LLMs. Writing for Teen Vogue, Anderson found that the tool increasingly performs the cerebral labor that was once expected of functional adults, both in the classroom and outside it.
One ChatGPT user admitted to using it to craft love notes to “a crush” and to introduce variety into the subjects of conversation she broached with friends. Another confessed to using it shirk the “burden” of drafting a heartfelt breakup message — one he presumably texted to his romantic partner. “Vivian” — all Anderson’s subjects were granted anonymity to discuss the taboo subject candidly — said she uses it to sound “less passive aggressive” and “angry” when litigating an argument with her roommate.
Some said that ChatGPT helps them organize and articulate their thoughts more comprehensibly. Others insisted that it helps them generate novel content for their work. Many talked about AI in therapeutic terms. “If I’m having a PTSD episode and I don’t have anyone to talk to, I’ll go on,” said one 21-year-old anxiety-sufferer. Another admitted to using it to avoid encountering the “negative/biased” advice that might be offered by friends. “Even if it is biased, it’ll be biased toward protecting you,” that student maintained. Some even talked about ChatGPT with the affection they’d typically reserve for human relationships. “I thought of ChatGPT like my little robot friend, as sad as that sounds,” one student disclosed. After all, “it can’t judge you or make assumptions.”
The piece does feature some flashes of enlightenment. One 22-year-old charity worker seemed sincere in her rejection of ChatGPT because it is “always so unbearably apparent” when her co-workers use it to draft emails. But most found value in it, particularly in its “mental health” applications. That’s not particularly reassuring. Given that the emerging consensus linking an epidemic of mental health crises among young adults to the amount of time young people spend interacting with online platforms, it’s unlikely that the cure for this outbreak of anxiety is even more internet.
The University of Chicago’s researchers concluded their study by noting that the social stigma surrounding the use of LLMs in an academic context will frustrate educational professionals who seek to integrate AI literacy into their coursework. That’s probably a necessary project. AI isn’t going away. Establishing best practices that govern its use and nonuse will soon enough become a core competency among students and professionals alike.
And yet that stigma cannot be dismissed as an unhealthy expression of thoughtless technophobia — not when we survey how young Americans are coming to rely on this thing to do their thinking for them, even for low-stakes engagements in social settings.