Ask HN: Are AI filters becoming stricter than society itself?
While experimenting with digital art and AI tools, I noticed how aggressively filters block historical, political, or artistic imagery. I wrote about how this impacts art, research, and cultural memory. Curious how others here see this balance between safety and censorship. https://tsevis.com/censorship-ai-and-the-war-on-context
It’s the diversity and inclusion slippery slope again.
The initial idea was good and very much needed to eliminate (or at least heavily reduce) long-established racism/bigotry.
But the problem is that a lot of people started to abuse it as a virtue-signalling mechanism and/or a way to justify their jobs, leading to insanities like renaming the Git “master” branch.
I suspect AI safety is the same. There’s a grain of truth and usefulness to it, but no AI safety person will intentionally declare “we figured out how to make models safe, my job here is done”, so they have to always push the envelope, even towards ridiculous levels.
Isn't political correctness just politeness ?
Many of the arguments against just seem to come down to "I want to be a jerk".
> Isn't political correctness just politeness ?
Maybe in theory, but in practice the cause has been subverted by both the professionally-offended who use offense to advance their own agenda or just seek attention, as well as those who want to virtue-signal by banning/shunning things that no reasonable person has a problem with. Both feed off each other while making things harder for reasonable people of any race/gender/sexual orientation.
> "I want to be a jerk"
I'd like to seriously understand who would be offended by the Git master branch being called "master" for example, especially since I don't believe the word in this context has any reference to slavery (but even if it did, I'd argue historical meanings of words are a good reminder to humanity of the horrible things we're capable of and to ensure we don't do those again).
No. Political correctness in practice is a way to exercise control by shaming people for adhering to long established norms of behavior and vocabulary.
Both of these are unreasonable generalizations. It depends completely on the purpose of the PC-ness, what groups are pushing for it, the cultural context etc.
I agree. For me, political correctness was originally an empathetic code of expression. I’ve never felt the need to provoke or ridicule another group of people. Of course, it did reach a point of hyperbole—where even historic artworks were misread as offensive or racist.
As the Greeks said: “Μέτρον ἄριστον”—balance is best. We don’t need to provoke just to indulge our worst instincts, but we also need tolerance for expressions that aren’t perfectly curated.
That said, my article isn’t really about political correctness. It’s about low-quality AI filters that can’t read context, and the corporate shortcuts that rely on them. My point isn’t to rant or complain, but to suggest a new era of content moderation—one that’s smarter, fairer, and more democratic.
FWIW: I’m Black.
Political correctness goes much further than politeness.
I was 46 when I entered BigTech (no longer there) in 2020. Of course the famous master vs main. But you were also chastised for saying “you guys” - even though the women said as much as the men, if you didn’t put the proper pronouns under your profile. I got chastised for saying “war room”. I heard an anecdote that now “pow wow” wasn’t allowed.
One podcast I listen to - Stuff You Should Know - replayed an old episode and added a disclaimer up front apologizing because the episode was about pregnancy complications and in the older episode they were being “heteronormative” by not calling out that this issue only affects cisgendered women.
I’m not saying anyone should be disrespected or I wouldn’t call someone by whatever name or pronoun they prefer.
I also found all of the “ally” groups, discussion of “micro aggressions” etc groan worthy.
This is in the DNCs charter
https://www.cnn.com/2018/08/25/politics/democrats-gender-non...
> The language now says committees “shall be as equally divided as practicable between men and women (determined by gender self-identification) meaning that the variance between men and women in the group cannot exceed one (1).” Gender nonbinary members will not be counted as either male or female, “and the remainder of the delegation shall be equally divided.”
Yes I would find this just as bad if the language was about race and someone of mixed race heritage.
I would much rather hang out with good live and let live traditional “guns and church family centered” old school Hank Hill conservatives (even though I’m not one), than the posturing “anti-racist” leftist wing.
I’m a White, Greek man, 58 years old, living in Southeastern Europe between Greece and Cyprus. Until my 40s, the world - at least in Europe - didn’t feel divided by trivial issues like “what counts as hate speech.” There was always a small racist minority, but the majority of people generally respected a politically correct language—not as oppression, but as a way not to harm others. Nobody felt “silenced” by avoiding harmful words. Then came the hyperboles and the rigid rules you describe. To me, these are very similar to the poor-quality AI filters I discuss in my article: they judge without context. But judging requires more than scanning words—it requires understanding how, why, when, and under what circumstances something is said. What you’ve experienced—and thank you for sharing—is the result of bad, simplistic policy created by people who haven’t truly tried to understand or adapt. Human beings, societies, even nature itself, are complex systems. They can’t be reduced to mechanical yes/no rules. That’s the heart of my article. And what I truly believe—as an eternal optimist—is that we can get to a place of smarter filters and smarter ways to coexist. It will take time, patience, and sincere effort. But it’s possible.
AI isn’t smart enough to permit edge-cases like artistic nudity, especially with people who’ll find and abuse any exemption for the edge-case to create something that doesn’t belong. AI is unreliable, so its censors are broad to minimize rare failures or bugs (“unintentional exemptions”) that people will find and abuse.
Despite this, AIs get fooled to this day. There are still jailbreaks for GPT-5 and nudity and piracy on YouTube.
The only way to distinguish “good” from “bad” is competence, which has never existed on a large scale.
Yes. For example, YouTube channels self-censor to an insane degree to avoid getting demonetized. Streamers don't dare saying fuck or shit in case the ai hears them.
True. The funny thing is, when I posted my article on r/artificialintelligence, the system automatically labeled it as NSFW. Proof that the filters themselves could use a little more intelligence. https://www.reddit.com/r/ArtificialInteligence/comments/1myi...
I think in some places AI's arent strict enough, its all an imbalance