ChatGPT Will Change Nothing
When you write publicly long enough, you eventually feel a strong desire to express a strong opinion. This is mine.
The development and success of ChatGPT, Dall-E, et al, will change nothing in our society. It should not be astonishing: we cheapened, offshored, then automated production of goods because we wanted so much of them; we now consume content and services in such great quantities that it makes sense to cheapen, offshore and then automate them, too. It is yet another stage in an already self-parasitic consumer society.
To toss my future self a lifesaver: I believe ChatGPT will make things worse and in doing so, technically change them. But it will effect change that was already in motion, ones that have been since some delicate equilibrium between media and human reality was disturbed in the late 20th century. American cultural critic Neil Postman would say the turning point was the development of the computer, the first truly inscrutable authority, and perhaps this tipped the scales from a mostly sensible and steady media environment to what we have today. At any rate, it is easier to imagine the world falling in love with ChatGPT in the nineties than it is in the sixties.
Now, we will have even more meaningless and false media; but we were already hungry for it. We will have even less beauty in the world; we already demanded endless novelty. We will have even less art; we already claimed it was too expensive. We will have even less thought; we already gravitated towards ease. ChatGPT will not define us. We defined it, in more ways than those inherent in a large language model. At at any rate, it will unemploy many folks that produce large volumes of uninspired content; but it did not change the culture calling for such a volume of uninspired content.
We may, for instance, be sure that soon no airline will pay for another palmy review of a destination, when ChatGPT and a prompt engineer can do so for a fraction of the cost, drawing on Google Maps and ten thousand Yelp reviews. Truthfully, I never enjoyed this tedious enjoyment of enjoyable settings, anyway (a dangerous thing for a president of a travel writers organization to say, I admit, but we must rise above this habit). Still, airline fliers, so accustomed, will still want to read these indulgent memoranda, and I suspect ChatGPT excels at them — and it is one less source of bread for writers who are conspiring to write better things. And imagine a future in which Google or Bing will provide you a praiseworthy and entirely tailored selection of love poetry for you and your partner, on demand (as well as ads for contraceptives). Poets and publishers of poets will be Donne for.
This decimation will be seen across many of the service industries, including (in what is the only delicious strain of irony in all this) engineers. ChatGPT, it is said, can code tolerably now. At the usual rate of advancement it should be CTO of the world in ten years and own a nice two-bedroom apartment in San Francisco’s Russian Hill. Say what you will about the foolishness of writers — and some of them are stupid enough to get English degrees — but they have never set out to write a book so good it renders all future writers redundant. It is almost anti-evolutionary.
I read a fulfillment of prophesy, in ChatGPT: we were told long ago by a wise man that "of making many books there is no end." Ditto for programs, ads, ebooks, product descriptions, scripts, poems: so long as the power to ChatGPT’s AWS servers stays on, I suspect the volume of text created in the coming year will outweigh all the millennia before it. The praisers of ChatGPT, tripping over themselves on LinkedIn to tell you how revolutionary this is, also say this is no different from other innovations, in that it will only displace one generation to the benefit of all subsequent ones. They do this so you will pay them to tell you how to get ahead of the curve and write good prompts for ChatGPT, before the service also provides this, too.
But they are wrong: nothing has threatened so much immediate and lingering harm since the atomic bomb. I think I preferred the bomb. It was candid. But as with the bomb, nothing foundational has changed. My assumptions for all the above is that ChatGPT and its kin will be used benignly, or at least not nefariously: we know this is not true and will find out how in the next presidential election. The worst features of man have gotten greatly exaggerated, and correspondingly our responses must be, too. Once ChatGPT does write its own prompts, the only remedy I can think of is a large electromagnetic pulse, or the Amish.
Perhaps there is a still better way. The most we can hope for from technology is that it can undo some previous innovation's unforeseen consequences. So I dream of a startup, fooled by promise of profit, into creating ChatGPT’s logical counterpart: an AudienceGPT, a consumer based modeled entirely off of what we like and getting better everyday. Pair the two and within milliseconds, they will have created the most perfect capitalistic cultural ecosystem possible, beyond understanding, and perhaps they can charge each other for the privilege, feeding each other banalities and hate, things that really go viral.
In the meanwhile, they will be too busy for us, and we will be forced to write some of our own content. Perhaps we will then realize some of what we ask for is not worth our time, either.
“Interior of Vaucanson’s Automatic Duck,” Scientific American
