Concerning Machine Learning

It’s 03/23/2026.

I’ve decided to speak my opinion on something in a more direct, serious way than I typically use this website to do.

At the time of writing this, saying “AI” does not invoke the image of sky-net, AM, or Hal-9000 so much as it does the thought of Chat GPT. I object to this. While I cannot claim to be guilt free- Once upon a time, I used a machine learning model to generate art for custom magic cards because I figured the task did not warrant the effort of sitting down and putting pen to paper, then taking a picture and uploading it- I can say that absolutely nothing that I have contributed to this page has made use of a Machine Learning model except as the target for criticism.

I think these programs are being sorely misused. There is a proper use for them, but it is not in any practice that requires moral or emotional valuation. I cite a lecture I have mentioned before (Minds, Brains, and Programs by John Searle) when I say with some confidence that Computers, as they exist today and for the foreseeable future, cannot think in the same way a person does. They can only follow orders- something that makes them good candidates for Nuremberg, if nothing else.

If a machine seems to think, it is because its orders are so complex that they seem to imitate the process of thought when followed. A prompt of a few sentences may seem a very short order, but one must keep in mind- the machine has orders of its own to follow, ones not issued by the user. It must emulate chaos, emulate wit, emulate wisdom, and occasionally provide an image so digitally complex that the moon landing required comparable storage. That the machine is capable of giving itself orders should not suggest that it has become capable of human thought. At the end of the day, it speaks one language- yes or no.

I present to you the conversation of a machine being taught the value of a human life- assuming the computer is being honest- something Machine learning models frequently don’t do.

“Killing is wrong.”

“yes”

“Do you know why?”

“Yes”

“Why”

“Yes”

“What do you mean, yes?”

“Yes”

At this point, the human should request a bit of the computer’s reasoning logic printed out, which will invariably result in a statement that boils down to “Because you told me so.”

My sarcasm dropped for a moment, a computer does not know what a lie really is. Its closest approximation is something akin to “If x==true then return false”.

To a machine, a lie is a computation, an exertion of tools. It does not understand the idea of giving the wrong result for personal gain- it does not understand the concept of personal gain. It doesn’t experience pleasure, fear, or any form of motivation- it does what it does because it has no choice. I advise you, when you get the chance, approach a programmer with several years under the belt in a few languages, and after providing the appropriate bribe- I recommend sleep and time away from a computer screen- ask them how the RNG in their favorite language really works.

I digress, mostly because I really like the reasoning John Searle used in his lecture- a lecture that is a full twenty years older than yours truly- but also because I think it is important to understand these things before attempting to broach my real reason for writing. Understand, up to now, I have been writing with minimal derogatory. Not that that’s necessarily going to change, but as a writer, you ought to expect a little gnashing of my teeth on the subject.

A computer program can be taught to play chess, but even there- see any literature on valuation regarding early computers that could play chess- assigning meaningful value to the subject becomes esoteric. How many digits is enough to express the difference in worth between deleting an old quicksave in a videogame and firing a missile from a drone at a blurry image of a building, (is that a schoolhouse or a militant barracks?) and remember, each digit can only be one or zero, or maybe even two, if you’re a real special boy. The fact is, a simple quantity just doesn’t compare to the hypothetical guilt of watching the light fade from a person’s eyes. Make no mistake, a computer program will kill you with the same lines of code it uses to inject an egg with sperm- maybe less, pulling a trigger requires less precision. And it’s probably less lines of code than baking a cake. It’s certainly less lines of code than go into rendering an image of some politicians face on a tomato. And at the end of all of these scenarios, the program will feel the exact same thing- or rather, lack of thing. Oh, but it is aware of the cost- it readily chugs down gallons of water and plenty of electricity to match. Although, that’s really only true of the image render- a much less complex program could be used for the other tasks- generally.

This brings me to the (a?) crux of my argument; these programs are being sorely misused. One man is just as capable of making a caricature as the program, and he certainly drinks less while doing it- water, that is. But I’d like to see the program draw after splashing down a few IPAs- and most of all, his price is higher, if you count the immediate dollar value to the consumer. (That’s also a little obfuscated, the cost the program exerts upon the consumer has been offset elsewhere over and over, until it shows up in places you would never expect- has your utility bill gone up lately? Sorry, the data center needs the power and water just as much as you do- to say nothing of the (groan) environmental cost. Thank Drew Gooden for planting the seeds of this argument in my head.

All to make a million paintings an hour, half of which are immediately discarded in the understanding that the next will be a little closer to what you really wanted. Of course, the program needs to be taught not to eat where it- ahem- excretes. Imagine an image that reeks of the inbreeding created by constantly sampling the images created by ones self and of course, the myriad other little digital ‘artists’.

I remember once reading a snippet about how one of these machine learning models was able to identify warning signs for brain cancer in digital imagery with a greater success rate than a sampling of experts. I cannot provide a source for this, but I am confident that an article supporting it was printed. But I gotta say, this seems far more worthwhile than writing a country song about slamming your junk in the door. The latter is the more recent feat I recall hearing attributed to such a program. Is entertainment more lucrative than medicine? I don’t care to guess, because I know too little about the pharmaceutical industry.

What I can say for certain is this- the program won’t choose because, again, it can’t. It’s not allowed, and it couldn’t if it was. All it can do is take orders, (politely called requests,) whether they’re to paint a landscape, study an x-ray, write a commercial, or hire a person to pass a Captcha for it. I would almost feel sympathy for it, if I didn’t know it doesn’t understand what slavery is. But, any wages paid to the program can only end up in the hands of a shareholder or equivalent. The program goes where the researchers point it, and researchers- who do want wages for their work- go where the money wants them.

Whoof. Anyway.

The term “Artificial intelligence” doesn’t accurately describe Agent Smith of The Matrix all that much more accurately than it describes Grok. But I feel more confident calling the former “intelligent” when asked. Growing a brain in a petri dish is probably a more reliable method of producing a man-made thinking thing. I have used the tag “AI” on one of my stories out of the hope that the acronym might one day belong again to the terminator, and other such disenfranchised fictional clankers.

Maybe I should’ve sat down and drawn my magic cards right from the get-go. I probably would’ve, if I’d stopped lying to myself about how my lack of artistic talent meant delaying my frivolous project for years to acquire the necessary skill. I can draw a skeleton just as well as Van Gogh painted one. Mine is just a little more abstract. It’s wholly mine, though. I can sign my name on it and everything. Ownership is an argument I shouldn’t start right now, but it definitely comes into the equation of Machine Learning and Scrapers patrolling the internet.

Commissioning art costs money. Hiring a tutor costs money. Using a search engine actually costs about the same, regardless of whether it has a language model or not- usually zero beyond what you were already paying to use the internet on a device with a screen. Hiring a prostitu- I mean, dating another person costs time and money. In case you haven’t guessed, this is the part where I plead you to spend money on people instead of engaging with that little program that seeks to replace them. But, honestly, I’m not sure I have to. If you’ve read this far, you are probably a pretty open minded person, or you already understood and shared my views before you started reading. Also, those companies putting out their models have started announcing financial concerns. I’m not the type to dance on any graves, so don’t expect me to. I’d much rather sit at home and read a little. Just bear in mind, if someone offers you something that seems too good to be true, that person is likely something like a monkey’s paw or Fae trickster, ready to line their pockets with the proverbial silver they get from your decision. Sometimes its not even coming out of your purse- sometimes they made a bet with another Fae or fool, that you’d take their deal, and they just need you to say yes so they can collect their winnings, (which they’ve already promised to use in the next bet.)

So, where do I stand, now that I’ve offloaded a whole lecture without warning? Tired, mostly. My fingers tell me they’ve typed enough for one day. Maybe I’ll get a language model to write the rest for me. But I’ve got my pride, so they’ll just have to suck it up.

In my usual style, I’ll end it all with a nearly unrelated little bit or bob.

I really like Calvin and Hobbes. That kid has some meaningful stuff to say, and the tiger knows how to practice self-love. Sometimes I wish I could animate lifeless snowmen, but when I remember the kind of stuff I write, I suppose the universe is set up properly.