Photo by Nikolas Noonan on Unsplash
Over the weekend, Bill Gates published an essay (“The turbulent AI era is here. The choices we make now are critical.”) covering his concerns about the development speed of AI products and the increasing need to regulate them. I was reminded of the story of the boy who cried wolf, only in this version, it’s the wolf crying wolf. Gates has been funding and encouraging AI products through Microsoft’s partnership with Open AI and others for more than five years. Those of us following the AI hype and press cycles aren’t surprised anymore when an AI leader “sounds the alarm” about the potential harm of tools they have been working on. It’s usually in the service of marketing AI companies by pretending to public-focused, or it’s a regulatory plea that’s actually focused at hampering a competitor who has gotten too far ahead. Relatedly, see Mark Zuckerberg’s essay “Future is for everyone: The Path to a Positive AI Future”, a piece of writing that tech journalists are calling a “manifesto.”
Let me take a pause here and recognize to myself and anyone reading: I went to graduate school to study and teach amazing literature. Here I am quoting “essays” by Gates, Zuckerberg, and (later this month) Dario Amodei — people I genuinely dislike and distrust. It’s not because I think these people are great or important writers; it’s because I find their words a revealing example of how out of touch they are with us whom they hope to be their users. Bill Gates is telling us that a turbulent era is here? He participated greatly in bringing us to “that era,” now he wants to make us responsible for the clean-up? Zuckerberg is writing about a future for everyone? Isn’t this the person who began his tech career by having his dorm bros rate pictures of women on campus? The same guy who later told us that privacy was a disappearing social norm?
We shouldn’t be believing these shills. The AI hype cycle, in whatever turns it takes, continues to fuel an industry that ignores safety concerns, privacy laws, surveillance norms, and general public good. When it comes to AI, at least the generative kind, it’s less important what they say and more important who they have shown themselves to be. If your paycheck gets signed by the AI industry, it’s likely you’re not going to be working in our best interest. And when you publish long “think pieces” on what we all should be doing about the products that “disruptive” products that you’re directly responsible for, I hope that you’ll understand our unwillingness to believe you.
Bottom line: even AI doomerism is AI advertising when it’s coming from a AI salesperson. It’s like when someone goes overboard and is screaming for help and when you reach for a life preserver to throw them, it dawns on you that their company makes that too (and probably the boat you bought a ticket on).
***
A recent article in The Atlantic (“A Society of Cheats” by John Paul Rollert) is worth reflecting on mainly because of the regressive nature of its argument about genAI attitudes and use in higher ed. It does what many professor-AI-response papers have done for the last four years: complain about students. The author’s unique contribution is to hang a critique of AI usage on the language of cheating, eroding morals, and cutting corners. He views his campus (and, by extension, all of ours) as the site of a “crime spree” due to students use of AI in their academic work.
Nevermind that state funding has eroded the support for the work we do with students, the security of the profession, the policing of what books we can read and ideas we can talk about; instead, the real crime is that students are using AI to cheat. This argument is frustrating on a host of levels, but I will describe two:
- The author is paying attention to the downstream affects of AI products when he should be looking upstream toward the corporations that built the products. Do I think that genAI products are horrible siphons of human thinking and creativity, and that we should be educating students at every turn about its negative impacts on the job markets, intellectual property, the environment, and equity concerns? Yes. Do I blame students for using a product that at least half of everyone around them is using, tinkering with, and explaining away? Not really. It’s up to us, especially those of us in the humanities who are more removed from the shareholder politics and business pressures, to make the case for how unsafe, unethical, and economically rapacious these products are. His argument comes off like blaming the sixth graders in the 1970s for smoking Camels. Our students are a vulnerable population; we should be getting in league with them and helping them, not dropping to the lowest rung on the ladder to lament their moral character.
- Look at the schools that the author mentions throughout his piece (Harvard, University of Chicago, Brown, Princeton); notice anything? It’s indicative of a certain class of academic privilege to make an argument about cheating. Educators in less “prestigious” universities understand all of these variables differently. Those students are working more outside of school, and chasing their diploma goal with a limited amount of time and resources and amidst growing debt. At more statistically “average” campuses (and usually at state run schools), faculty have been devalued and underfunded for decades so that more of us carry higher teaching loads, suffer precarity, and get paid less. At these schools we see harsher realities and more pressure. That doesn’t mean that students’ (or instructors’) use of genAI products isn’t concerning. It does mean that privileged people make arguments that, like it or not, ignore and then reveal that privilege. It’s too easy to go after the moral fiber of our students. Instead, we should be understanding these tools and the context in which they were made better. Regulation, refusal, pedagogical adaptation, political action, all of these focus the attention on the source of the problem: tech leaders who are disconnected, arrogant, and greedy. Complaining about student cheating is decades old. It won’t move the needle, and it just makes us look old and out of touch.
So, I’m annoyed at this article. I’m annoyed that The Atlantic printed it (haven’t we seen dozens of these pieces already in The Chronicle and Inside Higher Ed for years?). I share the perspective that genAI products are cheating tools, that they degrade our ability to connect with each other, express honest opinions, and treat each other fairly. I want people to stop using them; Rollert and I can agree on that. But let’s the blame the perpetrator and not so much the victim.
***
Lastly, in the “can this get any worse?” category (spoiler alert: yes), here’s an NYT article from yesterday: “Study AI Consciousness? The Bots Would Like a Word With You.” It’s a perfect blend of joyful and terrifying.