I occupy a rather obvious position in the spectrum of AI opinions. It’s evil and I hate it. This view isn’t a simple view per se. I have a great number of reasons for adopting it and decades of experiences that define evil for me. What is simple is how I deal with evil. Where I can, I oppose it. Where I cannot, I avoid it. If you subscribe to a moral framework that includes a concept of evil and you don’t count yourself among its legions, is there really any other alternative?

This is actually the most convenient thing about evil—if there is any convenience in evil anyway. If a person or an organization or even a concept leaves a place of nuance and relativity and crosses the boundary into evil, making your own decisions relative to it can benefit from this simplicity. Moreover, a religious principle I grew up with was this:

A good tree cannot bring forth evil fruit, neither can a corrupt tree bring forth good fruit. (Matthew 7:18)1

Understanding the tree is imperative to understanding evil. If you’re benefiting from or using corrupt fruit, as it were, and having to make justifications to yourself and others, it’s probably time to sit down with your mirror and have what my sister would call a “real talk.”

This whole business of using LLMs and so-called AI has never been a hard decision for me. I don’t touch them. They’re dangerous. They’re rotten. Their fruit is corrupt. All my decisions in regard to it is rooted in that understanding. It makes navigating the morality of it all very simple.

Just say no.

I’ve been doing a lot of reading (and YouTubing) on the subject as of late. I don’t know why precisely. It stresses me out. It angers me. It fills me with dread and doubt. To date though, absolutely nothing has done a very compelling job of presenting any of this as anything but evil which is oddly comforting.

Yes, Evil

I sense some people recoiling from my casual use of “evil.” In fact, there’s a very specific group of people out there I’m thinking of.

In this AI miasma there have always been the true believers. They either believe in the tech industry or just have some vague notion that new technology is good and will always trend toward our benefit. They believe this is going to like… cure cancer or something so no matter the cost, it’s like… totally worth it or something. (I guess.) These people are entirely delusional but at least they’re… delusional. It’s almost cute sometimes. Children are often delusional too and it can be very endearing.

These, by the way, are not the people I’m thinking of. They’re not recoiling at all. They’re just rolling their eyes at me. They know I just don’t understand that we’re gonna solve world hunger2 and faster than light travel in a couple weeks and I’ll be thanking Sam Altman for his service to humanity. Y’all are so cute. I should take you out for ice cream.

The group that I have come to find the most annoying and certainly the most complacent in the face of evil are the religious middle. What I mean by the religious middle are those who believe that there’s always some kind of middle space in matters of morality. I refer to them as religious due to the self-righteous smugness that comes from “considering all the options” and “keeping an open mind.”

Don’t get me wrong, the world is full of places where this is good and right and wonderful. The fact that this is often the right approach leads some people down the path that it is axiomatically always the right approach. I know people who adhere to this with a certain amount of zealotry—typically these types want to believe they’re some kind of intellectual. There’s an annoying piety in it all—a religion of “reasonability.”

Of course, not everything has a moral or technical middle ground. Some things are black or white, or good or evil. There’s very little room for debate and certainly none that’s interesting beyond philosophical masturbation. Some circumstances draw lines and you have to pick a side that is, by its nature, opposed to the other.

So, if you are someone who takes this religiously middle position about AI, what else do you take it on?

Is there some nice, comfortable, open-minded, reasonable middle for:

  • A flat earth?
  • The mass murder and cultural rape of the Native Americans?
  • Bathrooms and drinking fountains segregated by the spurious concept of race?
  • Child pornography?

Take your pick. And feel free to add some more examples. There’s no shortage of extreme example low-hanging fruit that can go into that list.

There are certainly people out there who are fine with any of the above. I have an uncle who unabashedly believes the destruction of the American natives was ordained by god and part of prophesy and as such, entirely acceptable because the fine white colonists are said god’s chosen. He’s wrong, of course. He’s absolutely wrong and I don’t waste any amount of time trying understand the middle between where he stands and where I stand. While I understand why he believes it—a completely broken moral framework—there’s nothing in the middle.

This brings us to our old friend, the modern LLM-driven AI industry and its applications and its unwitting defenders from the middle.

These are typically intelligent people. They understand that:

  • The economics are fundamentally unsustainable.
  • There is tangible environmental harm.
  • The reality doesn’t match the hype.
  • It’s being used by authoritarian business leaders to squeeze and intimidate workers.
  • The companies driving said hype are utterly dishonest and corrupt.
  • Slop and disinformation are growing like a cancer wherever there is media.
  • Security is a problem.
  • AI induced psychosis is a thing.

Do I need to add more bullet points here? How many of these do I need before it’s obvious there is no middle ground. The is no reasonable use. None. And yet…

People who understand all of the above continue to use this stuff and even help tell others about all the neat tricks it can do because, you know, piecing together UIs by chatbot is just so irresistible. Generating garbage documents from a couple prompts gets you through the work day. Talking to it like it’s a real person is super funny when it gets cheeky with you.

How utterly grotesque and evil does something have to be in order to just say no?

In short, I don’t give two shits about some “really awesome3” AI summary or search tool. Even if it was every bit good as you believe it is, can you justify its use in the face of what it is built upon? There are a lot of evils that are unavoidable. My life in the United States is effectively built on the parasitic exploitation of the poor and there is very little I can do about it. The very act of buying food feeds it.

The thing is, I need food.

What’s the need for playing with Claude Code at home?4

What’s the need for indulging these monsters at all?

So, for me, it’s easy. Do I need to really dig into Claude? Do I need to “get informed” about ChatGPT and understand what it’s actually good for? And look, I’m not making the argument at all that LLMs and the tools around them can’t do some impressive stuff5 and may even lead to gains in efficiency (maybe possibly sometimes). It doesn’t matter. The fruit is corrupt.

The companies building these models want everything and will use their wealth to coerce and legislate that they get it. They do not care who is harmed to get there. They do not care what is stolen. They do not care about the lives they destroy. They do not care if they tank the economies of nations because those at the top will have amassed enough wealth to be completely immune to the widespread destruction.

Moreover their leaders know they are lying! Sam Altman knows precisely where his economic games end! Dario Amodei knows exactly the width and breadth of harms to humanity his company means to cause! They know in their rotten fucking hearts and they they do not fucking care! They will sleep soundly while Rome burns around our children.

Please, don’t play with this shit and try to sell me on it. There is no such thing as a “reasonable” propagandist! Don’t slip in, “Yeah but… I did do this one neat thing.”

Yeah? Fuck you.

I have recently been listening the music of my late teens and early 20s. This invariably led me to Nine Inch Nails. As I listed to the Head Like a Hole I couldn’t help by think about how much it really summed up my own feelings on this:

God money, I’ll do anything for you
God money, just tell me what you want me to
God money, nail me up against the wall
God money, don’t want everything, he wants it all

God money’s not looking for the cure
God money’s not concerned about the sick among the pure
God money let’s go dancing on the backs of the bruised
God money’s not one to choose

Head like a hole
Black as your soul
I’d rather die than give you control

Remember where you’re dancing when you dance with OpenAI and Anthropic. There’s nothing reasonable about it. Don’t make quote Ruiner too.

  1. This is axiomatic. I get it. Calm yourself. ↩

  2. Not that we need AI to solve this. We have the food. We have the production. We have the money. Literally the only thing stopping this is a matter of distribution. ↩

  3. God. Is there any fucking work more overused among software nitwits than “awesome”? I have seen 30 foot waves hit lava spewed from a volcano in Hawaii. That shit is awesome. Your framework of choice is not awesome. ↩

  4. I full well understand some people do not have a professional choice thanks to the aforementioned evils of corporate authoritarians. While there is some degree of choice in where we work, there’s quite a chasm between those who use it with a corporate gun to their head and those who find any excuse to use it because they want the outputs. ↩

  5. Nondeterministically, of course. As a programmer though, typically want deterministic outcomes. ↩