I don’t care what the AI wrote
Robots are boring.
Embedded is your essential guide to what’s good on the internet, by Kate Lindsay and Nick Catucci.
ICYMI: Last week, I wrote a piece for The Atlantic about Instagram, and how everyone kind of hates it now. —Kate
A week ago, OpenAI released ChatGPT, a language model that “interacts in a conversational way.” My Twitter timeline—and probably yours—has not been the same since.
If you feed ChatGPT a prompt, it will spit back a surprisingly coherent response. My feed has become cluttered with screenshots of these texts exchanges, presented with awe and amusement and generally lots of hand-wringing about how this will change Writing As We Know It. Except all of this panic fails to take into account one very important point: I ain’t reading all that.
“my issue with the chatGPT posts is that no one actually reads big chunks of text that someone else screenshots and posts on here,” Atlantic writer Kaitlyn Tiffany accurately tweeted. “i'm just not going to read them no matter how clever your prompt was.”
I ain’t reading all that for the same reason I don’t read Lorem ipsum placeholder text: It’s not real, and for the most part, it does not say anything interesting. Terry Nguyen proved this point in a recent post for Dirt:
For the record, I tried to get ChatGPT to write this newsletter for me. I asked ChatGPT to write “a short first-person blurb about Al chatbots, specifically ChatGPT, from the point of view of a culture journalist who is skeptical about its ability to offer smart, stylistic writing.” I clarified that it should “offer at least one paragraph on its potential benefits and harms on society.” The result was… worse than mediocre.
As Nguyen points out, the sharing of these posts is more about people showing off their prompts than it is the text. The reason they’re interesting to anyone at all is due to what a human did, not a robot.
But I’m not being purposefully obtuse—I know the real panic isn’t about these posts, which immediately negate themselves by presenting the text in the (boring) context of being written by AI. Instead, the worry is what happens if and when people start turning to this tool instead of hiring real writers, or present AI-generated text in place of things like college admissions essays or cover letters. But will it replace actual, thoughtful writing? I’m not convinced.
My experience with anything remotely close to AI is relegated to NPCs in video games and my middle school conversations with AOL’s SmarterChild, but both appeal to the same human impulse: To fuck around until you find the boundaries of the technology. In the end, the NPCs repeat the same cycle of five stock sentences. SmarterChild would not insult me back. The moment you find out exactly where the robot falls short of the human, it becomes entirely uninteresting.
Technology’s come a long way since then, but it still has its limits, and as we find ChatGPT’s, its sheen will dull. Rarely has anything digital proven to be an all-out replacement for the IRL. We can all use Google Earth to virtually travel anywhere in the world—but when we want to see the Colosseum, we still get on plane. We still scream “REPRESENTATIVE” on the phone with Walgreens to be connected to an actual person. We abandoned the Zoom birthday party as soon as we were safely able. Humanity wins every time.