Never think alone

An AI rant? Well, yes, a bit of an AI rant.

I.

When the hand remains but the mind goes wandering, an animal can tell. My dog Pecan would thrust her nose at you accusingly if your attention trailed off while petting her, then grumble if she had to.

Another dog I once met would, his owners warned, nip at you if you idly extended a foot instead of a hand to rub the belly exposed at your feet. Give me the intelligent hand, the warm mind, or else leave me alone.

II.

Has it happened to you yet—receiving a fresh slab of AI-generated text as a substitute for communication?

I get this in my line of work from time to time. Usually in the form of “research” or “ideas” that do not, unlike human-generated research or ideas, reflect any principle of discernment on the part of the sender. What does the sender think about what they are sending? Well, they don’t. However much its marketing may claim that generative AI is only enhancing human capacity, not having to think is a large part of its point and its promise.

I haven’t figured out how to respond to these offerings yet. It’s a fool’s errand to engage with their substance, for there is no one on the other end to engage with. The sender is a medium channeling a mushy collective unconscious, serenely free of responsibility for the transmission’s content. If they have to refine or back up or assess the chatbot’s output, the promised time savings begin to unravel.

(And cutting out the human medium wouldn’t be much better. Anyone who’s spent even a little time with ChatGPT or its rivals knows how easily they fold when questioned. You’re right! My apologies.)

So far, I tend to not respond at all. And so far no one has circled back to ask me, “Hey, what about that idea ChatGPT had?”

These ersatz communications are offerings, though, presented in a generally helpful spirit, as if thinking were primarily a problem of quantity and here they come with a party-size container.

But words and ideas are cheap and plentiful. Always have been. The only hard part is making them good.

And this becomes harder still when you are confronted with an AI-generated void where the mind of a colleague ought to be. Behind a mirage of unmeant words whose sense slowly dissolves as you draw near, the person has slipped away.

III.

Is using AI in this way so different, though, from what we so often do at work? Fill in the forms, say the expected things, aim to broadly please?

Fair enough, to an extent. The self I bring to work is a somewhat muffled thing, odd corners and sharp edges at least loosely bubble-wrapped most of the time.

But if anything makes the work I do for hire good, it’s not the bubble wrap. It’s the hard thinking and real feeling I can’t help putting into it.

Part of this is just a personal (and sometimes regrettable) lack of aptitude for half-assing things. But however noble or inconsequential your work may be, I suspect it does damage the soul to absent yourself from it too far for too long.¹

IV.

“I think the true essence of what makes us human is our taste and our preferences and our sparks of inspiration and thoughts. That’s just supercharged by AI.”

—Roy Lee, founder of the “undetectable AI-powered assistant” Cluely, in a 404 Media interview

There is much to marvel at in this interview with a man who is right now being roundly dunked on for his vision of a world in which you might, after working all day at an ill-fitting tech job that AI has secretly helped you land, head out for a date during which AI feeds you lines and lies to use on the out-of-your-league woman across the table.

One thing that strikes me is how plainly Lee lays out the threadbare conception of humanity that underlies so many generative AI products. Our taste and our preferences: yes, at base, a human is a consumer. It’s the job of AI to remove us from the difficult and uncomfortable role of worker as quickly as possible and return us to our natural state of consumption: browsing through and enjoying the bountiful fruits of the AI’s output.

Sure, you can keep your “sparks of inspiration and thoughts,” which sound brief and pleasurable as cravings for ice cream. Beyond these flickers—ooh! I have an idea!—the AI guru finds no joy or value in the process of creation.

But whether you enjoy the process or not, it happens to be where you learn practically everything worth knowing. About the workings of the world, about other people, about yourself and your capacities.

It’s also where the actually good ideas come from, in my experience. The melody that floats into my head while I’m cooking dinner is a bit bland compared to the version I’ll work out later with a guitar. The script idea that works arrives five paragraphs deep into one that doesn’t really. As I begin to fill a blank page, I uncover associations that would have remained hidden forever had the idea remained in my head, or been documented in only the depth required for an AI prompt.

But even if quality were no issue, I do not think it is ultimately even pleasurable to make the process of creation less like work and more like shopping. When you’re browsing online and suddenly notice that you’ve scrolled through 12 pages of product listings and the sky outside has changed color—that’s a sickening feeling. Why make work or art feel more like that?

V.

“I will never have to remember when the American Revolution was. I will never have to remember what the capital of Wisconsin is,” Lee enthuses in the interview, as if bent low under the lifelong burden of orienting himself in time and space.

You’ve maybe had the experience of riding with a driver so dependent on GPS that they can’t navigate their hometown without it. What Lee proposes is: wouldn’t it be great to be like that driver in every aspect of your life? To have absolutely no mental map of the world at your command—just a collection of tastes, preferences, and fleeting thoughts to feed into the machine?

VI.

"We built Cluely so you never have to think alone again.”

—the Cluely “manifesto”

This is kind of a good line. Also a repulsive idea, if you think about it for more than a few seconds.

Personally, I actually do enjoy being guided by external forces in art and life—enjoy it more than most, even. I’m eager to exploit glitches, randomization, found objects, arbitrary constraints. Back when generative AI was newer and rawer (and before its oceanic appetite for energy and water was apparent) it seemed mildly useful to me in this way: as a weirdness-injection strategy.²

But generally, it’s bad to feel your head crowded by other voices all the time. I’m not only thinking of mental illness, though I do partly mean that.³ I’m thinking of the real or imagined judgments of bad bosses or bad parents, the drone of bad-faith commentary on worse news, the cobwebs of everything that’s been thought and tried and said before that obscure what may be possible. On top of all this, the AI guru invites you to deliberately pour the ultra-processed folk wisdom of the internet into your skull at every opportunity, and call it conversation.

Of course, a world in which everyone’s interactions are mediated by AI—in which you can never isolate the faint signal of a person’s intent from the chatter of their AI assistant—is a hauntingly lonely world. At least it is if you take an interest in life, in other people, in all the little static shocks that reality produces as the waking mind moves through it.

That most of us ultimately do take an interest has been my consoling assumption. Which should perhaps be downgraded to a hope.

1. Not that the line between self-preservation and self-alienation is always clear. And of course, if your work is evil, it damages the soul either way.

2. Here’s a sentiment you don’t hear as often as you might expect: “I was excited about AI and I’d love to use it, but I can’t anymore now that I know about the environmental impact.” Is this just because people generally end up doing what they want and consciences are easily assuaged? Probably partly. (And conversely, if you forego a pleasure for ethical reasons long enough, you may cease to identify with the self who once found it a pleasure.) But I also think that if environmental impact has much power to sway your actions in the first place, you are likely attuned to material reality in a way that precludes finding much lasting satisfaction in generative AI once the novelty wears off.

3. Where it becomes especially apparent that the voice thinking alongside you grows more dangerous the more it takes the tone of common sense.