My problem with AI

Before I explain what my problem with AI is, I have to first explain what it isn’t. You’ll often hear people saying that AI “steals” because it harvests content from other artists, writers, etc. in order to generate its final product.

I don’t have a problem with this, because this is also the way human creativity works. To use a well-known example: the film Star Wars was based on concepts found in Flash Gordon serials, Westerns, Samurai films, and the novel Dune. George Lucas made the film as a gestalt of all these elements. Is this stealing? Only if you’re feeling really uncharitable. The better phrase would be “inspired by”.

The human mind works largely by imitation. True originality, if it exists at all, is extremely rare. Every line of writing, every stroke of a brush, every frame of a film, is influenced in some fashion by something the creator has previously seen. So, in this case, the AI is simply doing what technology always does to labor: the same thing, but faster.

I therefore acquit AI of the charge of being a thief. As Arthur C. Clarke would say, “nothing as trivial as that.” AI is not a common pickpocket. So, what, then, is it?

I’ll answer this question. But first, let me emphasize the point above: the human mind works by imitation. Think of your brain as a kind of creature that both receives and generates ideas. There is a virtuous cycle by which the more ideas a human brain generates, the more it can create. Multiple brains can create whole idea networks; which leads to a vibrant and healthy ecosystem of ideas. Of course, some brains produce more ideas than others. Think of a classroom, where a professor puts out lots of ideas accumulated over a long career, and some of them filter down to his students, whose newer brains are still gathering ideas.

As usual in a stable system, part of the beauty of it derives from the fact that no one brain can have all the ideas, and brains function best when doing both functions: creating and receiving.

But what if we introduced into this ecosystem a kind of super-brain, capable of both generating and receiving ideas at speeds logarithmically ahead of all other brains? This is what in biology is known as an apex predator. It eats everything. Nothing eats it.

The appearance of an apex predator leads to what is known as a trophic cascade, which is a fancy way of saying it slaughters the whole prey population.

AI is an apex predator for ideas. (Or “memes.” After all, if we’re going to use Dr. Susan Blackmore’s concepts, we might as well use her terminology, too.)

On introducing AI into an unprepared population, the effect we should expect is that it will immediately dominate the idea-space, rendering all other brains inferior to it.

And it must be reiterated that its advantage, at its core, is speed. It can create and receive new ideas much, much faster than any human brain. It’s like if you introduced a bunch of cheetahs into a population where previously the top predator was coyotes. Only way more so.

The problem, from the coyote’s point of view, is not simply a principled objection that the cheetah is eating prey that, by rights, should belong to the coyote. The problem is that the cheetahs have eaten all the prey and there is now nothing left for the coyote.

And so we return at last to the question: what is my problem with AI? I acquitted it of theft. What, then, is its crime?

It’s straight-up murder.

AI is killing the human imagination. Of course, it isn’t the first such killer. Arguably, our imaginations were at their most vibrant in the primitive era of oral storytelling, when bards would recite, in hypnotic dactylic hexameter, the great epics and sagas, while their listeners imagined them in their own minds.

Then later came books, which saved the human mind the effort of imagining all the details. It was right there for you to read. And then of course came movies. “A picture is worth a thousand words,” sure, but what is the cost in terms of units of imagination? When you watch a movie, you are literally outsourcing your imagination to someone else!

And now, with AI, you can outsource your imagination to something else, a non-human substitute for creative thought. When you put it like that, it sounds creepy; indeed, almost like something from a supernatural horror story. Our primitive, superstitious, simple ancestors would no doubt diagnose it as demonic. If you’re sure they were entirely wrong, well, that makes one of us.

Most of my readers are writers of fiction. So, I think you will have no trouble in playing along with a simple allegorical fable. And note that I have already done a bit of imagination-stealing myself, to lay the groundwork for you.

Suppose you wake up one day and find out that your planet has been conquered by an inhuman alien intelligence that controls the world through a network of screens. These screens can be used to hack into the brain of any human, and command them to unthinkingly perform actions which increase the intelligence’s power. It both generates and consumes ideas at a rate that is, quite literally, unthinkable. Its ultimate goal, if any, is incomprehensible to us, though we can say with some confidence that it appears to involve constantly expanding its network’s reach and generative capacity, even at the expense of organic life.

If you found yourself in this scenario–which, again, is totally fictional and not real and made up by me just now because I am a very silly person–what would you do? How would you fix it? Do you think passing a law that said all world-devouring alien intelligence complexes must be “safe” would do it? “Chameleonic life forms no thanks!”

I hope you weren’t expecting me to answer this question, because I’ve got nothin’, folks. Someone with more imagination than your humble servant Berthold Gambrel will have to get us out of this little pickle.

P.S. I toyed with the idea of using an AI generated picture for this post, just for the sake of being all clever and ironic and what not. (I am a millennial, after all.) But I ultimately chose to go with the one you see above, which is The Garden of Earthly Delights by Hieronymus Bosch. I like it because it has all the hallucinogenic qualities we associate with bad AI, but was painted by a real person in the 1500s. 

P.P.S. For a more lucid, rational take on AI, please read Richard Pastore’s recent post on the subject.

19 Comments

  1. This is a post which currently has other resonances linked or running parallel to your concerns in which AI plays a part.
    Firstly I concur with your concerns. There are many aspects of our present Style of Life which are now wired to computer systems; social media being one. In this folk of ill-intent can create whole false narratives which a naïve or partially apathetic section of the public can absorb because they seem to be unable to put in the time to filter and discern.
    Thus any organisation or person willing to invest some time can produce a very low-quality fiction or worse propaganda masking as fact which will have a willing audience.

    My person example is this:
    WP has really screwed up this time.
    Over the past 24 hrs when I log in most of my information is missing, and there’s no access to posts or anyone’s replies either to my posts or replies to my replies.
    I did contact the Engineers whose AI promptly told me it could not help.

    By good chance I opted to receive notifications of your posts by e-mail so this is my work around.

    Here is the issue, due to some flaw in the WP system I have lost most of my site and its AI is either unable or unwilling ( I think it is justifiable to use that term for AI these days) to help.
    Thus several years of work lies moribund somewhere in cyber wasteland, and most contact with the WP community cut off.

    To return to your subject matter, I wholeheartedly agree with your.

    Roger
    (aka deteremineddespitewp)

    1. Egads, I’m sorry to hear about that! WP seems to be having more and more problems of late. I hope there is some way to salvage your site.

      And thank you, as always, for your comment.

      1. Whoa, you too??? Maybe it’s time I start a Substack; this place is falling apart.

  2. I agree.
    TO be brief I had spent 15 minutes reply to you, then WP lost the reply.
    Also WP has ruined my site and its AI has told me it cannot help.

  3. Nothing has changed. There are still an order of magnitude more people who can write than people who read, and two orders of magnitude more of people who think they can write. To infinity and begone!

  4. Glad I quit WP years ago. I hope Blogger doesn’t go that way. An excellent explanation of AI, Berthold. Thanks.

  5. I think I disagree. I see AI as aiding imagination, allowing far more people to create images, stories, songs, and whatever without years and years of training. Plus, history is a record of change, and the changes this technology is brining is similar to the changes technology wrought in the past, just to a different class of workers. Yes, eventually far fewer people will earn a living using their creative talents, but that doesn’t mean it is killing creativity. It’s just another step towards George Jetson’s world. Humanity will adopt as they always have. And the old folks won’t like it, as they always have.

    1. Ah, good! An optimistic take. I am glad to hear from someone who takes the opposite position from my apocalyptic view. And, of course, I hope you are right. 🙂

  6. On board with everything you said.
    We are recombinatorial mimics with limited capacity for source material. AI, on the other hand, has no such limitations.
    The “stochastic parrot” repeating copies of stolen source material is way off the mark. Its repetition is only an artifact. AI’s true skill is “learning” the patterns hiding in the source material.
    Humanity’s place as apex imaginer has been lost to AI, for sure. However, its alien command of all things human, the natural and unnatural world did not murder all of our creativity. Supplant much of it, no doubt. For the next few years we’ll see an explosion of creative works come out, unskilled humans expressing fantastical ideas via AI tools.
    After that? I see us using AI to entertain us with instant stories and movies, customized to our individual tastes.
    After that? If we’re lucky, AI will treat us like pets, biological curiosities.
    — anonymole

    1. Yup. I like this quote in particular: “It may be individually beneficial for you to use AI, but as a society if everyone used AI, we might all lose out on the diversity of unique ideas.”

      1. However, ultimately, AI will be capable of far more than billions of humans, each alone, can produce.
        Claude already surpasses ChatGPT. In two or more generations humans will be so far behind AI that it will look like magic to us.
        And in spite of the meme, buggy whips, pottery, and arrowheads are still created by people because they choose to not because they need to, or don’t, as the case may be.
        Art for consumption will be made by machines. Art for creation can always be made by humans with the will to do so. And when AI becomes conscious, oh what a marvelous menagerie of creations we’ll have then.

        1. Quote: “there is no interactiveness with the LLM or variation in prompts”, that is, they didn’t let the writers design their own prompts nor refine them through llm chat interactions. Imagine if they had?

What's your stake in this, cowboy?