Dave Eggers accepted an invitation from OpenAI's chief executive, Sam Altman, to speak at the company, and used the occasion to tell the audience that its technology is doing damage to young writers.
"There is no safe amount of AI in humanities for kids under 22, at least in writing," Eggers said, according to an interview he gave the Irish Times. He put the case for what is lost in blunt terms: "If a young person decides to shop out their voice to a machine, then they become voiceless."
The claim he made to the room
Eggers told the OpenAI audience: "You're adding 20 hours a week to the workload of every public school teacher in this country and the effect is apocalyptic on young people being able to develop their own voice."
That figure is his own assertion rather than a measured finding, and he did not present supporting data for it. The underlying complaint, that teachers are spending substantial additional time trying to work out what their students actually wrote, is one that teaching unions and individual educators have raised repeatedly since generative AI became widely available.
The phrase most associated with his argument came from a separate appearance, on NPR's "Wild Card with Rachel Martin" in June, where he described "an entire generation tempted and too many of them acquiescing to the silencing of their own voice in favor of a bland, unthinking machine."
Why he thinks writing is the wrong thing to automate
Eggers founded the youth writing nonprofit 826 Valencia and co-founded the publisher McSweeney's, and his objection is less about cheating than about what writing is for. In his account, a rough draft and a disjointed thought are not defects to be smoothed away; they are the process by which a person works out what they think.
He has been similarly dismissive of AI-generated fiction. After Altman promoted a machine-written short story, Eggers said the output was "pastiche garbage that would fool only the most gullible," and that what such systems do is "mimic syntax," the San Francisco Standard reported.
He is not a newcomer to the subject. His 2013 novel "The Circle" imagined a technology company whose pursuit of total transparency became a form of control.
The other side of the argument
Eggers's position is contested, including by many teachers who are not employed by technology companies.
The case for AI in classrooms rests on access and workload: that a student without a tutor at home can get immediate feedback on a draft, that teachers can generate practice material and differentiate work for mixed-ability classes, and that students with dyslexia or writing in a second language gain a scaffold they would otherwise lack. Schools and universities have adopted the tools widely, and the direction of travel in most systems has been toward regulated use rather than prohibition.
There is also a historical argument, that calculators, spellcheckers and word processors each drew warnings about lost faculties and were absorbed anyway. The counter to that, which Eggers's supporters make, is that those tools automated transcription and arithmetic rather than the composition of an argument.
What is not yet known
The empirical question underneath the dispute is unsettled. Evidence on whether AI assistance helps or harms the development of writing ability is early, mostly short-term, and often produced by parties with an interest in the answer. Studies pointing in both directions exist; none yet follows a cohort long enough to say what happens to a writer who learned with these tools from the start.
That is roughly the position Eggers is arguing from: not that the evidence is in, but that the experiment is being run on a single cohort of students, and that the cost of being wrong falls on them.
