How I Use AI, And Where I Cheerfully Refuse
Here’s my AI policy.
Okay, let's get this out of the way first: nothing you're reading here was written by A.I., cause that would be gross. All of my life I have been a wordsmith and grammarian, even spending eight years of my life as an English teacher… and yet of late I have found myself inserting informalities, improprieties, and lo! even minor errors! into my writing in order to distinguish it from the structurally flawless drivel that many people are passing off as their own words.
Onward!
The first thing I want you to know, dear reader, is that I am against human replacement and for human creativity, agency, dignity, and even silliness. My favorite moments as a reader are when I am completely surprised by the turn that a story or an argument takes… which is a sensation that I never get when reading LLM outputs, but often get when reading GK Chesterton or GM Hopkins.
My second principle re: AI is that I flatly reject the rhetorical devices of inevitability and doom. Far too much of the discourse around AI has breathlessly assumed that its ubiquity is inevitable, and therefore you should incorporate it into everything immediately in order to not get left behind, but that also it's so powerful that it could quite possibly cause the end of human civilization in the next three to five years. These are PR/marketing plays at best.
AI is, of course, already changing the labor market, but in different ways and in different degrees from field to field. But to paraphrase Mark Twain, reports of humanity’s death have been greatly exaggerated. Like Bartleby the scrivener, you are perfectly free to say with regard to a particular use of AI, or regarding AI in general, “I prefer not to.” And like the choice to forgo a smartphone or delete a social media account, such choices may come with inconveniences, but who said efficiency was the highest value?
OK, cheerful rant over. With the philosophical prolegomena out of the way, here’s how I’m am using - and not using - AI in my work:
I am using it to automate some of the mind-numbing administrata of running a small business, like categorizing financial transactions, organizing files, reformatting data, etc.
I’m not using it to make design or strategy decisions. If you work with me, I hope it’s because you want my expertise in your corner… and that’s what you get.
I am using an AI-powered transcription tool for meeting notes and dictation. The notetaking helps me stay even more fully present in my conversations with people, not having to interrupt their thought or mine in order to scribble it down. And although I'm a pretty fast typist, voice dictation is way faster.
I’m not using it to generate images from scratch. There's tons of decent stock photos out there already, but the best way to build trust with your website is always to use real photos of you and your place and your clients or customers experiencing success. And AI-generated images are, conversely one of the quickest trust destroyers for a lot of people. Plus they look creepy - they give me the uncanny feeling that you sometimes get when looking at a lifelike doll. So close, but so gross.
I am using it to edit images, like removing a distracting background objects or upscaling a low-res image. But I still do plenty of image editing by hand, too. I'm proficient enough in Photoshop that I can quickly knock out many of the kinds of edits that people struggle to get AI image editors to do.
I’m not using it to pretend to be me. Some folks have AI writing all their emails now, with some even letting it respond without their review. Other folks have AI programs doing cold outreach, scraping publicly available info about you to try to write an email that looks like it was written by someone who is interested in you. Those are an abomination. If you get an email from me, it's actually from me. (Although there are sometimes scammers pretending to be me, so always double-check the email address.)
I am using it to write code, or to debug code I've written, to build out my designs. The designs come from the 3-pound brain inside my thick skull, and are still built largely by hand, but where the platform in question is limited, I use a few different AI powered tools to create custom code solutions.
Why am I using AI for some things and not for others? Looking back at the list above, the line I'm drawing seems to be between tasks and relationships. Insofar as it involves other humans and/or my embodied expertise and creativity, I will happily do it myself. If it's a solitary task, and if it's more technical or rote than creative or open, I'm willing to consider handing it off.
A coda: the thing that most concerns me about AI is the ways that the reigning models and chat-based interfaces seem to be hacking our attachment, in the same way that social media and Web 2.0 were (and are) hacking our attention. The folks at the Center for Humane Technology have done some really clear thinking on this. One of the rules we have in our house (my wife and I are raising three girls) is that we don't pretend machines are people... because one of the unfortunate side effects of doing so is that we may begin to treat people (including ourselves) more and more like machines. And that's not how I, at least, want to live.
“It is easy for me to imagine that the next great division of the world will be between people who wish to live as creatures and people who wish to live as machines.”
-Wendell Berry, may he rest in peace
