By I. Pliskin
In the past couple days, OpenAI has released its “4o” image generator. With it, the company has had many users of its software flooding social media sites with computer generated images in the style of anime studio “Studio Ghibli”. It’s worth noting that the CEO of OpenAI, Sam Altman, stated on xwitter that the company puts “a lot of thought into the initial examples we show when we introduce new technology.” In this case, it seems the company has chosen to show its ability to plagiarize the stylistic elements of the highly acclaimed animation studio in a bid to make a definitive statement about their new software. The image they’re trying to paint isn’t so much the ink and gouache of animation cells. OpenAI is trying to paint an image in people’s minds that it has surpassed its mechanical limitations, and can now easily match its output to the height of human craft and creativity.
Yet when we peel back the curtain we see that the Sam Altmans of the world aren’t “tech wizards” as much as they are “wizards of oz” so to speak: pushing and pulling at knobs and levers to ultimately produce what amounts to little more than a series of illusions. In a big way, our current society is built on the backbone of the things we think we see, but in actuality aren’t quite as substantial as they appear to be. I’m decades late to saying that the “American Dream” is a tall tale marketed to people through sitcoms and advertisements. At least back in the day we could dream of one day being able to buy a home made out of solid stuff like bricks, concrete, and hardwood. Nowadays, you need to amass a small fortune to put a down payment on a house that’s 90% plastic and petroleum bindings. Home ownership isn’t the sole hope and dream here – the houses themselves are built and held together mostly with hopes and dreams!
That’s not exceptional, that’s just how our entire economy is now. Elizabeth Holmes was able to keep her Theranos scam going as long as she was because the healthcare system we have was already designed from the ground up to only appear helpful to a large amount of people, not to actually help them. That’s why no one cares about that one health insurance CEO who died. Private health insurance isn’t designed to help people who need treatment be able to afford it – it’s designed to take rents from people under the illusion that it will cover surprise costs if necessary. Yet it was found that the insurance company that had its CEO walked down knowingly denied patients with a system that had a 90% error rate. The rub is that this system used a machine learning algorithm akin to and probably based on the ones that OpenAI makes.
In all, the big story here isn’t that OpenAI has crossed over some threshold of human creativity and intelligence. Their plagiarism machine still can’t handle well established issues with their previous generators like drawing a hand that has 5 and only 5 fingers. I guarantee their new algorithm would shit itself trying to generate something like a single scene of unique animation style like “The Amazing World of Gumball” because its output is still based on generating an output that is an average value between a prompt and the massive amount of data it’s stealing and plagiarizing. The tool hasn’t learned anything new since its last version – they just put more data in for it to average out. It doesn’t “know” how to “draw” anything with a particular “style” because it isn’t doing any of those things. The big story is that companies like OpenAI really want the general public to believe that these tools can draw, can think, can imagine.
We have a pretend economy being propped up by people pretending like they’ve taught servers and computers how to pretend. Does that make sense? Even some of the more tangible parts of our economy like construction are based in cutting a lot of corners that look fine at an initial glance but that are likely to break down very quickly and at the slightest disturbance. A big thing that supplements that is a rich amount of media that allows many to disassociate from their actual living conditions and project themselves into the fake lives and realities they see on screen. The big push for “AI” represents not just further corner cutting into tangible things, but an attempt to cut major corners in how media is made which people like to escape into. This double corner cutting has bled into the making and selling of the “AI” tools themselves. The makers of these tools are doing a lot to try to show these tools off as being a lot more effective than they actually are, but even if the end results are as good the makers of these tools want you to think they are – they remain so unprofitable and ecologically destructive that it still wouldn’t make any sense to use them in they way they’re being marketed.
The message in all this? Don’t believe the hype. And I’m not just talking about “AI” machine learning algorithm generative whatevers. We’ve got to take a more critical stance on our lives and circumstances overall. The reason insurers decided to put up a tool that just denies everyone whether their claim is legitimate or not is because they knew ahead of time that most people wouldn’t question the denial and wouldn’t push back. Police have been known to do all sorts of witness and evidence tampering to paint images of guilt onto the innocent. Your landlord painted all the walls before he leased you your apartment to cover up the mold that was growing there. Individually and collectively we need to develop a growing awareness of these illusions being set before us, that cloud the way to us fighting for real and substantial improvements in our day-to-day lives and in our overall conditions.
The Philadelphia Partisan 