AI Slop Sucks, But The Idealized Alternative Is Not Honest
A submission sarcastically titled Please Use AI front-paged on Hacker News a week ago.
It’s an interesting bit of poetry. I don’t want to make too much of the specific examples it presents, but use it as a kicking-off point because much of the anti-AI rhetoric uses these sorts of high-contrast dichotomies as their narrative foundation.
A sort of “the alternative to someone using AI generated slop is performing the most personal, heartfelt, involved, considered, thoughtful alternative, with full care, concern and attention”.
Is that really the experience of anyone, anywhere?
Coding, for instance, given that’s where I’ve spent my career: I can unequivocally say that Claude Code / Opus 4.8 generates solutions better than 99%+ of the code I’ve come across over my career, including important, lauded projects. Dramatically better code1.
More secure code. More “considered” code. Better commented code. Higher performance code. With a better separation of concerns with better support for future extensibility.
Could good developers make code at the same level or better with the proper effort and consideration? Of course! But often we’re throwing together something as quickly as possible to solve an immediate problem, don’t want to bother “premature optimizing” design or performance considerations until it reaches a critical problem level, etc.
The vast bulk of code on this planet is terrible, and it’s important that people are honest about this. It is sub-slop. It is insecure, poorly considered trash, and this is why enterprises having a large code footprint is very seldom an asset, and often is instead just a massive debt that grows with every line added.
This same observation holds across almost every field. Massive industries of people generating thoughtless, lazy trash.
The zero-creativity, copy-paste advertisement world, for instance, where the same ads are regurgitated countless times, just mixing up the product and demographics for what you’re targeting. For instance the classic silly, stupid husband does something dumb, and his rational, intelligent wife makes an “oh gosh” look at the camera and saves the day. I instantly despise any product that has this sort of trash ad, which means that I carry a lot of emotions for a lot of product given that it’s by far the most common advertisement template.
A large percentage of television and movies now are just remixed updates of stuff that was successful in decades past, albeit maybe with the creative direction to “put a chick in it and make her gay”. An entire field where the vast majority of content is worthless slop long before AI, armies of shameless “creatives” just doing the laziest possible thing.
This same thing can be seen everywhere. Wedding speeches are just a cliché. The whole idea of “genres” in art and music is an admission of follow-the-leader remix and repeat behaviour, sometimes taken to such hilarious extremes it seems like parody. And if people really think most kids wrote creative, thoughtful essays before LLMs came and corrupted the youth … were these same people born yesterday?
Sure, a small, tiny subset of students did, but the dominant, average behaviour was just cliché, maximum-lazy slop. The dictionary defines cliché as “1 : a trite phrase or expression”…
Large volumes of creative and intellectual output has been slop for decades.
Is someone individual if they put on the “goth uniform”, looking precisely like the genre of “goth” is supposed to look? Is it rebellious and tough to jack up your Dodge Ram and hang truck nuts from it and roll coal, just like the countless other “rebellious” guys pursuing exactly the same image and following an identical template? Or to buy a Harley, grow a beard and apply the “tough biker” template?
It goes on and on. There is perilously little that isn’t already lazy “slop”. There are entire industries where almost all of the participants are just slop purveyors, long before they asked ChatGPT to slop it up for them.
Most of us are just following a script and playing a role. Much of humanity has long been robotically stumbling through the motions with minimal effort.
I linked a video from someone who put it much better than I can previously, and that video is worth a purposeful, attentive watch from beginning to end. Don’t ask Gemini to summarize it, or just jump around because you want to get back to Reels. Just play it through and actually listen. It’s spot on.
This Isn’t A Defence of AI
There were a series of terrible Apple ads back when they were trying to pretend to be adept at AI. Guy gets AI to lengthen and aggrandize a message to his boss. Boss summarizes that message with AI, and then writes a response that he again lengthens and aggrandizes with AI. Someone else feigns familiarity with someone in a social situation by using AI to supplant actual familiarity.
This ad was so horrendous I have to think the Apple executives that okayed it got punted from the company, because anyone using AI like that is a fool wasting everyone’s time.
Similarly, evaluating resumes with AI, where the resumes have been created with AI, and then a response is sent with AI, and your openclaw reads the response, and… What are we even doing? What is the point of all of this nonsense?
Vibe coding on the whole is largely fantasy, and making real solutions still requires an enormous depth of technical mastery and knowledge, particularly when you encounter the edges.
Not to mention that there are enormous, dystopian dangers with AI that we need to seriously confront, for which we need strong governments rather than relying upon the benevolence of plutocrats. These dangers are incredibly pressing, and risk upsetting the power balance so completely we’ll quickly find ourselves at the point of no return.
But head-in-the-sand denialism or false dichotomies are so easily and obviously nonsense that they don’t delay the onslaught for even a microsecond. They’re just pablum that denialists use to assuage each other.
On The Explosion In Fantasy Data Centres
Trillions of imaginary dollars are flooding into the AI space, and everyone is trying to get a cut of the action. This has given rise to an enormous industry of fantasy data centres, many with preposterous proportions and imaginary attributes. Many of them are by developers having no credibility in the tech industry, no known users of said data centres lined up, but they’re just proposing galaxy-sized developments and hoping that proposing big square buildings is enough to cash in on some of that AI valuation pop.
Many if not most of these data centres will never be built. And of those that do get built, many will end up abandoned.
The epicentre of fantasy data centre grifts has to be Alberta, where a whole array of pretend projects have been proposed, grifters lining up with hands out hoping they can loot investment funds, governments, and so on2. Alberta, you should know, has a derelict, obsolete power grid, and some of the highest electricity prices in Canada. It has no geographical benefit either from a population proximity or geological perspective.
Footnotes
-
This is where someone attempts to elevate themselves, declaring that only bad programmers think current AI makes good code. These people are idiots, and their noisy nonsense — basically a head-stuck-in-sand hope that you can invent a reality and make it real if you just say it enough — can be seen in every discussion about AI. It’s obnoxious, boorish nonsense, and their noise has been tolerated for too long ↩
-
Remember when Jeff “Hands Out for Welfare” Bezos made a bunch of cities do a pathetic campaign to petition for “HQ2”, promising some megopolis if only the taxpayers of those cities give his outrageously profitable company tax benefits? NYC has been demonstrated to be absolutely correct in telling him to get stuffed, and the pathetic “HQ-little-2” in Virginia, now stunted and sad, just betrays everyone that danced for that beggar3 as a fool ↩
-
I mean, if Bezos was bad with that HQ2 trash, sports leagues like the NFL4 are a million times worse. It’s simply incredible that cities are still dumb enough to give hundreds of millions to billions to beg the NFL to please build a stadium and keep all the profits. That this scam still works to loot taxpayers is amazing. Los Angeles was such an exception with SoFi, happily watching beggar teams leave when they couldn’t get their handouts, then come crawling back and magically find the ability to pay for their own stadium. LA gave $0 for SoFi stadium, unlike the many other pathetic D-tier cities that loot the working class for the billionaires ↩
-
The Olympics and FIFA and other leagues are also guilty of this scam. This nonsense is always pitched with super-sized claimed benefits that never come to fruition, and with undersized-costs that end up massively exploding, again for taxpayers to foot ↩