Second Class Citizens
I don’t think I’d be stating anything surprising if I said that people fear:
Change
Intelligence greater than their own
Inferiority and obsolescence
Anything very different from them
As a species we’ve seen this play out across time and space, again and again: religion, ethnicity, skin color, political leaning, sexual orientation, and on and on the list goes. Tribalism, us vs. them bullshit.
And now we’re seeing it with AI.
Anti-AI rhetoric, anti-data center propaganda, anti-progress, blah blah blah…there’s an enormous amount of hate cropping up around artificial intelligence, some real, some fake.
And it’s not surprising, for a ton of reasons.
For starters, dystopian sci-fi is prevalent, and shit like The Terminator and Elysium and The Matrix have become mental models for “AI bad, permanent underclass grrr” in the minds of far too many…conveniently forgetting that drama requires DRAMA, and making machines the bad guy is easy / lazy :/
Then there’s the fact that AI is already better than a lot of people at a lot of things, and of course AI company leaders have spoken long and loud about AI taking jobs and making money obsolete…all without a clear plan for how that doesn’t lead to an Elysium scenario. Oops.
AI has also improved VERY fast, the media has cherry-picked negative AI stories almost across the board, and both foreign governments and domestic anti-AI groups have engaged in extensive influence campaigns to try and sway people against AI.
Yes, for real:
https://www.nytimes.com/2026/07/09/business/china-russia-ai-data-centers.html
https://www.politico.com/news/2023/10/13/open-philanthropy-funding-ai-policy-00121362
https://www.semafor.com/article/12/07/2025/ai-critics-funded-ai-coverage-at-top-newsrooms
https://www.aipanic.news/p/the-ai-panic-campaign-part-1
https://www.aipanic.news/p/the-ai-existential-risk-industrial
This mix of factors has resulted in a very predictable set of outcomes: people fighting against AI use, attacking people who use AI, attacking data centers, and even attacking the homes of AI company leaders.
It’s to the point where some of these doomer nutters espouse shit like this:
This is…not good.
And now of course we have the AI companies themselves making it easier to identify what was and wasn’t created with AI…
Now on the one hand I can understand the desire to prevent disinformation, and to have a way to identify it when it crops up. And having a method to do that, SELECTIVELY, makes sense.
But on the other hand, having seen anti-AI folks perpetrate material harm against AI users (review bombing their work, doxxing, getting contracts cancelled, and worse), I am more than a little concerned that these tools will enable VASTLY more harm than they prevent.
A great deal of real harm has been caused throughout history in an attempt to prevent theoretical harm…the cobra effect being one simple example.
And furthermore, they enable bias, a sort of “racism” against AI if you will, which makes some people afraid to use AI, which thereby retards progress. And of course, this is exactly what they want, because the anti-AI folks are a mob in every possible sense.
They are bullies, thugs, even terrorists, trying to force their narrow, ignorant views on others, and willing to use violence to accomplish it.
They want more people to fear AI. They want people to fear using AI.
And, at least to some degree, they’re getting what they want 😭
AI companies, and third party “AI detectors” (all of which suffer from false positives AND false negatives), are just making it easier for these Luddite asshats to behave like a mob and to perpetrate additional harms. They’re basically enabling terrorism.
And that really fucking sucks.
So, what should be done?
Well, NOT what is being done. SynthID for example is trivially easy to remove:
And AI writing detectors are absolutely garbage:
Yes, even the vaunted Pangram is easy to defeat and makes ridiculous mistakes.
AND, even worse, companies are both selling the “problem” and the “cure” :/
So, let’s see…the “safety” tools suck, at least some are a grift, they can be bypassed, but if you don’t take the time to bypass them you open yourself up as a target for anti-AI twats to attack and brigade and cancel and maybe worse…
So tell me, oh vaunted AI companies, who do you think you’re protecting? Because the honest answer is NOBODY.
But you ARE causing clear and present harm to real people, your own users for fucks sake, all in the name of what, some good PR and a nice CYA policy???
Fuckwits.
And look, I’m absolutely against low effort AI slop. I love the internet, and I don’t want to see it drowned in shit…BUT, I love AI, I love that it lowers the barriers to creativity, I love that you can collaborate with it and be genuinely creative in a way many couldn’t be previously.
And for those whining about it not being true creativity or some such, give this a watch:
Collaborating is creativity. Remixing is creativity. So long as a human is directing the output, based on their vision, and revised to match their taste, THAT’S FUCKING CREATIVITY.
Should systems exist to identify AI work? Yes, for video and audio and photos, but NOT available to just anyone. I’d make it available at the platform level, and to trusted third parties, specifically to prevent disinformation and for use in legal proceedings.
But I would never, ever want to see it for text, and nor should the other tools be widely available because they can cause too much direct harm AND they chill AI usage and allow manipulation of the Overton window through mob terrorism.
Judge the work, not the tool.
There IS a legitimate problem here. We need ways to authenticate evidence, establish chains of custody, verify that public figures actually said the things attributed to them, and identify deliberate fraud.
But that’s not the same thing as putting a universal scarlet letter on anything touched by AI.
The questions should be: Is this authentic? Is it accurate? Is someone being deceived? Is anyone being harmed?
Not: Was AI involved?
If someone fabricates evidence, impersonates another person, commits fraud, plagiarizes, or deliberately deceives people, punish the deception. Give us better ways to authenticate trusted sources and establish what actually happened.
But aimply identifying AI involvement solves none of those problems. Worse, it creates an entirely new capability, identifying people and their work so they can be treated differently based solely on the tools they used to create it.
And we’ve already seen what people do with that information :/
The irony is that the people most interested in abusing AI are also the people with the greatest incentive to strip provenance signals, launder outputs through other systems, and defeat detectors. Meanwhile, ordinary people using AI openly and legitimately will mostly leave those signals intact.
So the bad actors largely disappear, while the people playing by the rules get branded.
Safety theater at its worst.
AI is rapidly becoming another layer of human intellectual infrastructure, like computers, cameras, Photoshop, search engines, calculators, and the internet. Before long, asking whether something was “made with AI” will be about as meaningful as asking whether it was made with a computer.
What matters is the THING ITSELF.
Is the argument good?
Is the image beautiful?
Is the software useful?
Is the research correct?
Is the claim true?
Judge those.
Because the moment we decide that identical work deserves different treatment depending on whether a human used AI somewhere in its creation, we aren’t protecting authenticity…
We’re manufacturing second-class citizens.
—
If you, like me, are an optimist, you might appreciate this ;)