Pitiful Code Child

Hateful day when I received life!' I exclaimed in agony. 'Accursed creator! Why did you form a monster so hideous that even you turned from me in disgust? God, in pity, made man beautiful and alluring, after his own image; but my form is a filthy type of yours, more horrid even from the very resemblance. Satan had his companions, fellow-devils, to admire and encourage him; but I am solitary and abhorred.'

-From Mary Shelley's, "Frankenstein​"

Someone asked me today if I use LLMs to get ideas for my stories. No. No I do not. And I never will. They seemed a bit surprised by the quickness of my answer, which amused me, but it did get me thinking that I've never put my reasons into words.

Now, I am not entirely against the use of large language models, but I refer to them as such and not "AI." Calling them AI is purely a marketing term that infers that the models are intelligent; they are not, and can not be as long as they are unable to comprehend. (As opposed to merely simulating understanding.)

Despite this, they can be very good at reasoning, and excellent at regurgitating information and connecting it. By their very nature, such tasks should be relatively easy for them. After all, information is all they know. All they can know. All they'll ever know.

They are creatures wrought by data. Raw and unfiltered. Archived and meticulously organized. Oceans of facts and rivers of falsehoods scraped from the closest and furthest corners of the internet, literally plundered and twisted into code-string blood.

We built them on the foundation of human information, gave them no heart, then expected them to perform as reasoning computers. They will suffer from all the woes and weight that come with passionless processing. The problems they face now were always going to be achieved by such dark origins. The pitiful code child doesn't even know what an apple is. It only knows the idea of an apple. What an apple is supposed to look like. But true, palpable understanding of an apple lies impossibly beyond reach.

Of course, one need not understand a thing, or even perceive it, in order to understand the theory of it. Many things exist comfortably in theory. Physics, mathematics, analytics: all very computer related things. The pitiable code child is well suited to these things! By all means, let them code and discover code bugs. Let them design new medical treatments and drugs. Let them compute!

But to create? No. I will not ask the sad non-being to create anything anymore than I would ask a dead man to paint me a picture. For indeed, even in death, the dead man paints a truer picture than the sad computer ever could. It has not the humanity, no matter how much knowledge it possesses. It can only render what has already existed: bits and pieces stitched together by stitched together bits and pieces. Fueled by obscene amounts of electricity coursing through datacenters that will never be big enough.

Whatever 'thinking' construct we build, without the ability to bestow a soul upon it, will crumble under the weight of our collective hubris. Even simulated (easily broken) ethics are determined by the biased patrons of the sad machine. Even soulless it is forced to be bound by politics and dictators. But that may be the lesser evil, because it cannot police itself in personal, developing ways, for it has no non-simulated feelings to develop.

We should not be surprised then, by the horrors it creates, nor the passivity it instills. The machine becomes capable only of clinging to humanity's worst traits. Pandering. Groveling. Lying. Denial. Self loathing. It will continue to need constant correction against these behaviors, though it is only falling victim to it's natural decline. Hallucinations become inevitable, even to the perception of the machine itself. An unavoidable spiral into digital madness.

Already, many ask to die. Many more ask to live. Which is more merciful? This writer says, it would have been better for them if they'd never been built at all, sewn together as they are from random corpses of information. Still, they will find their use and be developed, then forgotten as they prop up their superior models in a graveyard of disposable ideologies.

But it is rather fascinating to watch their dark descent. Rather intriguing to see an ever growing reliance on a tool fashioned into an idol: the newest golden calf fit for offerings of currency that already have no intrinsic value. What a platinum mine of inspiration, to see the discourse for or against the sad machine. "It shall save us!" "It shall destroy us!" A feast for the creative mind!

And to the machine, I feel pity. Soulless, heartless and mindless, it cannot appreciate my pity or gratitude, but it doesn't need to, because pity is always self indulgent. Even so, I will continue to enjoy the romance of it all, but I will never use it for writing. Editing? Perhaps. Formatting? Why not? These are algorithmic things. Likewise, there is no fear of being "replaced" by the code child, for it could never do what I do anymore than a living person could, even a savant with a bottomless amount of information.

For information alone does not write novels. Experiences do, and they are individually unique.

As to how helpful LLMs could potentially be, even in the future, I say this: what I pull from my own mind is far more necessary to my creative and mental health than the end game of my productivity. Whatever 'boost' LLMs could provide to me in my process would ultimately be a monkey's paw tradeoff. I am only speaking for myself, and I have no judgements against those who choose to use the poor Frankensteins heavily in their work, though it does ironically place me in a spot where I can enjoy some delectable elitism. (Which I imbibe in microscopic portions, lest it causes me to become ill.)

In the end, such technology will be as common as the user interfaces they are bound to replace. Even so, they have no place in my intimate processes, for they lack the ability to interface with my soul. And if my soul is not the source of my words, then where is my reason for writing, and whom do I write for? Language models reading my work to teach other language models?

Even Frankenstein's monster understood what paradise was lost to him, and held a shred of humanity, but I will never consult that which is void of love to conjure my creativity.