One of the most salient features of our culture is that there is so much bullshit. Everyone knows this. Each of us contributes his share. But we tend to take the situation for granted.
Harry Frankfurt
There are shit jobs, and there are bullshit jobs.
I'll let the late great American anthropologist David Graeber explain:
a bullshit job is a form of paid employment that is so completely pointless, unnecessary, or pernicious that even the employee cannot justify its existence even though, as part of the conditions of employment, the employee feels obliged to pretend that this is not the case.
Bullshit jobs are unjustifiable. Academic administrators trumpet 2030 plans to try to justify their jobs’ existence, but no one is fooled. Corporate lawyers don’t even bother arguing that their job makes the world a better place. These jobs are bullshit, through and through.
But they’re very well-paid. Upper-level administrators make several times what faculty make, and dozens of times what I'm paid as a graduate instructor.
And that’s strange, because if pay is supposed to reward hard work, it’s going to all the wrong people.
At a university, no one works harder than the janitors. Their jobs aren’t bullshit at all; these workers are essential. But they aren’t paid well, and their jobs are bad for them—difficult, exhausting, mind-numbing, even degrading. Janitors have what Graeber calls shit jobs.
In shit jobs, you’re paid and treated poorly.
Graeber notes that as the United States has moved from a manufacturing to a service economy, there’s been an explosion of bullshit jobs that don’t seem to do much besides keeping people employed. Even at many non-bullshit jobs, the bullshit quotient has increased over time. Nurses report having to babysit so much paperwork and technology that they have little face-to-face time with patients.
Now introduce ChatGPT.
Say you need to turn a short thought to your boss's boss into an important email. You ask ChatGPT to expound on it at some length. On the other side, your boss's boss receives your impressive-looking report and decides it’s way too long to read. So they ask ChatGPT to write them a handy summary. Then they tell it to dress up their own reply in jargon suitable to their station and line of work.
Like zipping and unzipping files back and forth.
I’m all for automating away shit jobs. (Without creating a new miserable class of AI slaves—that would just hand shit jobs off like a hot potato.)
But I’m very worried that by increasing our capacity to produce text by at least one factor of magnitude, ChatGPT will proliferate bullshit from sea to shining sea.
To test my hypothesis, I had ChatGPT write me a bullshit email.
And then we edited it for a while.
I’ve talked about Frankfurt’s notion of bullshit before, which is essentially speech or action without appropriate concern for truth. ChatGPT-3.5 could already bullshit, but ChatGPT-4 spews better, more human bullshit.
People already can’t tell what’s written by ChatGPT and what isn’t, and neither can software. It was sort of possible with ChatGPT-3.5, but you’re really screwed with ChatGPT-4, especially as people learn to iterate on ChatGPT’s first draft and write with it as a collaborator.
Last week, I presented my idea that we should teach students to write with ChatGPT at an online conference. A professor excitedly accused some of my text of being “obviously mechanical.”
But I'd written it from scratch.
So people already can’t tell what is and isn’t written by AI, and they’ll only get worse at it from here. At some point, it might become reasonable to assume that most text you read is probably AI-generated.
So are we doomed to drown in a tsunami of bullshit?
Doomed is strong. But bullshit multiplies, often in ways we don't intend or notice.
Say you’re a teacher who still insists that everyone write their assignments by hand, and keeps trying to play Whack-a-Mole with plagiarizers. You're confident that you can tell who's writing with ChatGPT and who isn't, and even which parts of their writing are most suspect.
What is your job like a year from now, when ChatGPT-4.5 is out?
Well, you might still catch a few kids who weren't clever enough to go back-and-forth with ChatGPT (or couldn't pay for the latest version) and happened to turn in a wonky first draft as a result. But will that still be the case for ChatGPT-5 or -6?
Are you sure?
And look, some kids might still end up learning something by talking to ChatGPT about their essay, but if you aren’t designing at least some of your writing assignments to work with ChatGPT, you’re failing to guide your students' interactions with a powerful piece of technology whose accuracy they cannot verify.
As a philosophy instructor, I can tell when ChatGPT makes mistakes—about philosophy. But in basically any field where I’m a non-expert, I'm underqualified. The best I can do is start Googling.
Fortunately, everything on the internet is true.
If we don’t train students to become more thoughtful non-experts, then when they inevitably do turn to ChatGPT to do their homework, they won’t know how to critically evaluate AI writing. That means we'll be doing them a huge disservice in a world where reading and writing with ChatGPT may well become less and less optional.
You’re still teaching as though they’re staring at a blank page holding a pencil. But instead, maybe you should teach them querying and fact-checking and editing before you get them drafting?
ChatGPT raises the floor of text production: Everyone can now produce texts of at least this quality in one minute flat. And the floor is only going up from here.
As ChatGPT gets better, students may not even bother talking with it for long. They’ll shoot off the first draft straight to you (maybe even trained on phone logs of their own speech patterns) and you’ll love it.
And then you'll wake up one day to find yourself assigning an AI busywork and living in denial. At some point, you went from English teacher to bullshit worker—and you brought your whole class with you.
So yeah, I think we need to teach students how to write with ChatGPT, as well as how to write from scratch. But none of us are really trained for that. So I get being scared, and not knowing what to do.
Well, maybe you can start by asking ChatGPT for suggestions?
Subject: Our 2030 Plan: Empowering Students and Shaping the Future of [University Name] Together
Dear [University Name] Community,
I hope this email finds you in good health and high spirits. Today, I am pleased to share an overview of our 2030 Plan, developed by our administrative leaders and faculty while taking into account the valuable perspectives of our student body. This plan focuses on empowering students and shaping the future of our esteemed institution, ensuring that student voices have been considered throughout the planning process. Our plan revolves around 8 key pillars designed to strengthen and support various aspects of our university's operations.
And don't worry, we've even considered easing the parking situation! (Just a little humor to lighten the mood.)
Here are the 8 key pillars of our 2030 Plan, with input from our students:
Student Well-being: Foster a supportive environment that enables our students to thrive academically and personally.
Innovative Curricula: Develop cutting-edge programs that prepare our students for future challenges and opportunities.
Research Excellence: Strengthen our research capabilities and foster collaborations that drive innovation and discovery.
Global Partnerships: Expand our network of partnerships with leading institutions worldwide, promoting knowledge exchange and collaboration.
Infrastructure Development: Invest in state-of-the-art facilities that enhance the learning experience and support our community's needs.
Faculty Development: Attract and retain top-tier faculty by providing professional development opportunities and competitive compensation packages.
Sustainability and Social Responsibility: Cultivate a culture of environmental stewardship and community engagement across our campus.
Financial Stability: Ensure the long-term financial health of our institution through efficient resource management and diversified revenue streams.
In the coming weeks, we will host a series of town hall meetings and information sessions to provide more details on the 2030 Plan and its implementation. We encourage all members of our community to participate and contribute to shaping the future of [University Name]. Further information on these sessions will be shared in due course.
Together, let us embrace this vision for the future and reaffirm our commitment to excellence, innovation, and student success.
Best regards,
[Your Name]
Provost
[University Name]
Good news, comedians: Your job is still safe.
We’ll see how you fare against ChatGPT-4.5.