You don't have to use AI
stop it right now
Yesterday, I was at the gym, struggling to squat properly and sweatily listening to Ezra Klein interview an Anthropic executive, when it occurred to me: this stuff is so, so much worse than even the most dramatic doomers have made it out to be. The stakes are existential. Turn it off!!!!!!!!! I beg you, TURN! IT! OFF! I would honestly prefer for a sentient AI to kill every last human being on the planet in some hideously gruesome way than for even one more of us to become the kind of amoral, thoughtless person-shaped vacancy that AI threatens to turn us all into. At that point, I would welcome a violent singularity. Better to lose to the machines than to become the machines.
For the past few years, I’ve tried not to think about AI too much, and shockingly, I’ve succeeded. It’s been more than half a decade since I’ve taught at a university (although I hope to do so again someday), so I haven’t had to scheme to keep my students from outsourcing the exquisite work of thinking to a glorified search engine—and AI is worse than useless for a critic, so I haven’t been the least bit tempted to use it in my line of work. I’m not really one for optimizing away the various inconveniences in my daily life, preferring to luxuriate in inefficiencies, so I haven’t turned to it for scheduling or the like.
Of course, it’s long seemed obvious to me (and, I presume, basically everyone I care about or respect ) that the use of AI in most humanistic endeavors is beyond the pale. Even if it were good at writing prose or doing philosophy—and thus far it isn’t —to use AI to write or philosophize would be to render those activities futile. In some of the sciences, some of the time, the point is the outcome—the vaccine, the medicine, the finding, the technology. In the humanities, the point is the process. The point of writing is to make something beautiful or interesting; the point of reading a book or a philosophy paper is, at least in large part, to make contact with another human mind that has strained to make something beautiful or interesting, whether or not the human mind has succeeded or failed. There would be no reason to read a novel that an AI had written, even if AI got to a point where it could write “well,” as perhaps it will if it keeps cannibalizing the greatest fruits of human endeavor with impunity. Writing a book with AI would be like driving a car to a marathon finish line, then claiming the title. Even considering using it to write evinces a complete misunderstanding of the enterprise. I can’t stress it enough: if these claims are not obvious to you, you are not “in my world and not of my flesh,” to paraphrase a remark of Stanley Cavell’s in “Aesthetic Problems of Modern Philosophy.” I hope you are ashamed of yourself and your dwindling humanity, and I think it should be legally required for people to say when/if they’ve used AI in the process of writing something so that I can avoid it at all costs.
All of this was clear to me at once, but I had never had occasion to think about the use of AI for thoughtless, menial tasks, because I never really felt much pull to use it, and I don’t really know anyone who does. I was sort of agnostic about using AI for dumb things. I didn’t use it often myself, but I have used it for two purposes: when I was on a low-iodine diet as part of my thyroid cancer treatment, I made AI generate low-iodine recipes for me (there’s iodine in next to every kind of food, and the charmingly clunky recipe book that the Thyroid Cancer Association distributes is nearly unusable for vegetarians), and I periodically ask it which medicines I can take, given my unwieldy spate of diseases and prescriptions. (NB: if you have an inflammatory bowel disease, as I do, you should try to avoid Ibuprofen.)
But after listening to the Klein episode, I’ve decided that even using AI for minor, silly tasks is eroding our moral and agential faculties to an untenable degree. I’m not saying that you should never use it for anything—I believe the quants who tell me that it has applications in STEM fields I’ll never want to understand, and I can imagine reasonable if highly limited uses when it comes to busywork like bibliography-construction—but I am saying that the bar for using AI for anything, even something stupid and mindless, should be extraordinarily high.
What struck me about the Klein episode is that Jack Clark, with his Anthropic bona fides, is supposed to be one of the good guys. And in fairness, I don’t think he said anything outright evil. What he did do was display the extent to which his moral imagination has been warped and degraded beyond what I can accept in a fellow human being.
At some point in the interview, Klein asks him about the pitfalls of engaging with relentlessly affirmative and ingratiating chatbots—Klein, by the way, does a very good job of holding Clark to account throughout—and Clark replies that AI needn’t be used for self-affirmation. In fact, he explains, he uses it to help him occupy other people’s perspectives: “I’ve used these A.I. systems to basically say: Hey, I’m in conflict with someone at Anthropic. I’m really annoyed. Could you ask me some questions about that person and how they’re feeling to try to help me better think about the world from their perspective?”
He didn’t even pause before confessing that he’d used AI in this way and didn’t appear to think it was a cause for shame or contrition. aaaaggghhhHHHHHHHHHH!!!!!!!!!!!!!!?? I don’t know what’s worse, using AI in this fashion, or thinking that using AI in this fashion is such a normal, acceptable, and forgivable thing to do that you should admit to having done so on a podcast in a national newspaper. Imagine having a fight with a colleague, friend, lover, etc, and learning that the colleague, friend, or lover with whom you have a human relationship, whom you trust to approach you in human terms, is outsourcing the ethical labor of relating to you charitably to a machine???? I mean, I would simply leave my husband on the spot if he ever did this. No questions, no divorce filings. I would open the door, grab my dog, put on my running shoes, and sprint as far as possible in the other direction. What a despicable violation of the basic ethical contract linking one person to another!
What brings a person to the point where it seems natural to consult an AI for help with the most basic functioning of the moral apparatus? I’m sure in Clark’s case there are a lot of factors. For one thing, he works at Anthropic, and I imagine his coworkers do not have a normal emotional response upon hearing that he’s taken their interpersonal conflicts to Claude. But I also fear that using AI regularly alters a person’s instincts and habits of mind and heart. When you’ve gotten used to forgoing all the little frictions of thought, all the pleasures and annoyances of grappling, then even the ethical imagination can come seem like the sort of thing you may as well “streamline.” Sure, maybe once or twice a year you can use AI to do some especially odious practical task without thereby transforming yourself into a monster. But remember what you are at risk of losing each time you do, and never do so lightly. Don’t use AI to get recipes; don’t use it instead of Google. There is a difference between reading a text that a human being typed out and reading one that a machine excreted. It’s good for you to know that you are making contact with a bored intern when you read a blog post. It’s good for you to let your mind snag on the difficulties and the basic boredoms. It reminds you that annoyance is part of what makes you a person. And, for now at least, you remain a person. You can afford to let the chatbots learn to empathize from each other, secure in the knowledge that human beings have done so without their help for millennia.