Recently I received an AI-written email from a friend. It wasn’t sent to test AI, or to show it off, as in “ha ha check this out”; my friend had a question to ask me, and the email asked it over the course of a few paragraphs. It then disclosed that, oh by the way, I used AI to write this. My reaction to this surprised me: I was repelled, as if digital anthrax had poured out of the app. I’m trying to figure out why.
Career-wise, I’m fortunate in that I don’t need to think about AI very much. While I work for a tech company, it’s one that—against stereotype—doesn’t chase trends and fads, so “AI for business” just hasn’t come up in my everyday life. On a personal level, I’d describe myself as fundamentally uninterested in the kind of AI technology that’s in the news today. I’m a designer and I like designing things; I don’t think I need or want or would at all appreciate a computer doing the creative stuff for me. If Adobe adds better selection tools that happen to use AI, right on; but as for it putting ideas in my head, that’s off-limits. I’ve got plenty of my own ideas, and if I ever need new ones, I can draw on the wonderful world around us, with its long and rich history. That sounds like a more fruitful approach to me, and it also seems more fun than typing into a command prompt.
Where exactly would I draw the line between helpful features (“make this red shirt green instead”) and offensive takeovers (“generate an album cover in the style of barney bubbles, award-winning”)? As I said, until this email I was more bored than enraged by AI, so I didn’t have an immediate answer. I use computer crap all the time—it’s pretty cool! So what was different here? I thought I’d come up with some comparisons that capture different aspects of my friend’s AI email, in order to see how I feel about them.
- Getting an AI-generated email from a friend didn’t feel like they had used autocorrect to type better.
- It also didn’t feel like they had used phrase-level completions, e.g. one-tapping to turn “in my hu…” into “in my humble opinion.”
- It didn’t even feel like Grammarly or a similar service was employed to fix errors of syntax and fumbles of style. None of these tools would change what was being said, or even meaningfully influence how it was done.
- That cheesy sig saying “Sent from my iPhone” that people rolled their eyes at in the 2010s? That appears goofy now—but this didn’t feel anything like that either. That was a Daria sticker placed in the corner of the page; it wasn’t the totality of the letter.
- It wasn't like Oblique Strategies or another creativity juicer—those give you a tiny and strange starting point, and you do the bulk of the work. They prompt you.
- What this did feel like was, it felt as if my friend had buzzed their secretary over the intercom and barked at them to send me a letter, signed “R. Jeeves on behalf of ——————.”
- It felt like getting a birthday card with only the prewritten message inside, and no added well-wishes from the wisher’s own pen. An item off the shelf, paid for and handed over, transaction complete.
- The email felt like getting a form letter, one of many thousands sent out by a large agency; except it was sent to me by, you know, a friend.
- It felt like the episode of Mrs. Maisel where Midge discovers that her husband’s comedy act features stolen Bob Newhart jokes.
- It felt like a family fridge decorated with printed stock art of children’s drawings.
- It felt like opening the front door at my birthday party to welcome in a group of iPads on wheels instead of people I like.
- It was like using a phone tree to kiss.
I knew that I didn’t want an algorithm to design layouts and draw illustrations “so I don’t have to,” but prior to this email, I never even pondered whether I wanted AI to call me up on behalf of people in my life. It had simply not occurred to me—and now that it has occurred to me, I definitely do not want small talk and relationships outsourced to server farms. This stuff shouldn’t feel hard or taxing; it’s what our presence here on Earth is mostly made up of. The effort, the clumsiness, and the time invested are where humanity is stored.
I don't know why my friend sent this the way they did. Perhaps they did in fact just want to test it out. (It didn't feel great to be tested on.) Maybe they thought this was just the new normal. Maybe it's my fault for being known as a Tech Guy. Surely I'd love it!
The AI email didn’t capture my friend’s tone or mannerisms—I imagine the LLM they used had nothing to go off of but a brief prompt. The AI did, however, try to sound like someone. It was folksy and upbeat, talky and pretend-excited (a computer can’t be really excited). It referred to itself using the human word “I,” as in “I was also wondering about…” It thought it was people.
Years from now, could an AI that was trained on all of my friend’s emails and texts and personal documents sound convincingly like them? Could it be so advanced that I wouldn’t even be able to tell that my friend hadn’t written to me at all? Possibly. And that idea saddens me the most.