🐱 YouTube's algorithm blocked channel, deciding conversations about chess pieces were "hate speech"

  • 🏰 The Fediverse is up. If you know, you know.
  • Want to keep track of this thread?
    Accounts can bookmark posts, watch threads for updates, and jump back to where you stopped reading.
    Create account
CatParty


YouTube’s haywire automated censorship systems removed a chess video after interpreting chess language as “hate speech.” Curious about the removal, a group of researchers tested AI software, similar to what YouTube uses; their experiment resulted in more than 80% of comments on chess videos being flagged for hate speech.

Last summer, YouTube removed a video on a popular chess channel owned by Antonio Radic, a Croatian chess expert. The Google-owned platform reinstated the video 24 hours later after Radic complained but did not explain why the video was taken down in the first place.

But Radic, also known as Agadmator, speculated the reason for the blocking was the use of chess language. Phrases such as “black,” “white,” “threat,” and “attack” are popular in chess. These same phrases could easily be misinterpreted by a censorship AI as so-called hate speech.

In the video that was flagged, Radic was discussing chess with Grandmaster Hikaru Nakamura, the youngest American to achieve that title and five-time winner. According to computer science experts at Carnegie Mellon, the use of the phrase “black vs white” might be the reason YouTube’s AI flagged the video for hate speech.

Following the incident, the Carnegie Mellon researchers decided to conduct an experiment. They run more than 680,000 comments on chess channels through an AI programmed to detect “hate speech.” Their test resulted in more than 80% of the comments being lagged for hate speech, mostly because of the use of “black” and “white,” which, in chess, refer to the two opposing sides.

While the researchers did not use the same exact AI system YouTube uses, their experiment does show there is a need for social media platforms to train their moderation AIs to recognize chess language.

YouTube did not provide a reason why the video was mistakenly flagged. In a statement, a YouTube spokesperson for the platform said:

“When it’s brought to our attention that a video has been removed mistakenly, we act quickly to reinstate it. We also offer uploaders the ability to appeal removals and will re-review the content. Agadmator appealed the removal, and we quickly reinstated the video.”

The researchers noted that if such censorship is happening to a channel as popular as Radic’s, it could be happening to many other smaller chess channels.

Radic’s channel, started in 2017, has more than one million subscribers, and is considered the most popular chess channel on YouTube. It’s most popular video, where Radic analyzes a 1962 match between Oleg Cherinkov and Rashid Nexhmetdinov, has more than 5.5 million views.
 
Click to expand...
CatParty said:
While the researchers did not use the same exact AI system YouTube uses, their experiment does show there is a need for social media platforms to train their moderation AIs to recognize chess language.
Click to expand...
Or you could just get rid of the AI bots all together. They don't solve problems, they only make them worse.

I'd hate to see Chess go "progressive", as it's a really fun game.
 
SSF2T Old User said:
Or you could just get rid of the AI bots all together. They don't solve problems, they only make them worse.

I'd hate to see Chess go "progressive", as it's a really fun game.
Click to expand...
Problem is get ready for the barrage of lawsuits from (((music groups))) because youtoob didn't shut down those nasty vids about how awful our songs are or the ones that use it for 5 seconds
 
I find it strange that using the words white, black, and attack in the same sentence was enough to get flagged, but the 50 million videos where some woke activist talks about how the white underclass is literally genociding black and brown bodies and that's why we need to exterminate the pale ones doesn't.

Though I would also be completely unsurprised to learn that they programmed their AI to suddenly specifically ignore anyone with skin darker than a half and half.
 
NJBear said:
Chess is also sexist as the Queen can move anywhere while the King gets one step. And the bishop is anti Catholic too! And the Knight and Castle are a dog whistle for White Supremacy. The Horse is a victim of Animal cruelty forced into war. And those Pawns...
Click to expand...
Castles are homophobic as they can only move in straight lines. Knights are harmful to players' self-esteem as their movement patterns are associated with taking L's.
 
Honestly I can sort of understand how this might happen given a shitty AI programmed by incompetent pajeets. What's a bigger mystery to me is why every now and then my recommended feed used to get absolutely fucking rammed with this dude's vids, despite my utter lack of interest in anything chess related.
 
I suspect pissing off chess nerds who have access to presumably better algorithms and software and have been designing them purely for fun since the 1980s is going to end badly for Google.

Generally a lot of chess nerds also make their careers in high level computer programming or engineering as well. I'm rather confident that anything these people can build will outclass Alphabet Inc's feeble AIs right out of the gate and the technology gap will only get worse with subsequent iterations.
 
"While the researchers did not use the same exact AI system YouTube uses, their experiment does show there is a need for social media platforms to train their moderation AIs to recognize chess language."


No, there's a need to END the tyranny of intentionally-biased "AI", it's not AI, it's a program that's been told what words are no-good by dangerhair programmers. We don't HAVE true AI, and all it's mistakes are not mistakes, but unintended consequences of people deliberately trying to censor you for politics, not for "hate speech".

After a decade of being worked on, and hyped up as something great by the whole tech community who spend their treasure on "improving" it still is a half-baked shambles. In the end, it only blocks cat videos because it thought the cat with a black spot on it's chin was Hitler and annoys innocent people who say "black" in casual conversation, what's the point? It hasn't made one detectible reduction in the level of hate in society, if anything, it's pissed off people even more that their benign stuff gets flagged by the robo-Nanny, but obvious calls to violence by the Black Panthers is let through... which is a silver lining I guess, showing how BAD it is shows that it's no accident, they're TRYING to come up with a way to make any unapproved race-talk just disappear, but keep getting outed when the robot does the job TOO WELL.

Censorship "AI" is a farce that's getting harder and harder for tech to play dumb with.... it's why they don't even answer most questions about why you got banned - they don't want to admit the system is a hot mess of inconsistency because it was programed by biased idiots to limit political discourse.

They need to be called to account for this and not GIVEN a chance to improve it, but told to turn it the FUCK off.
 
You must log in or register to reply here.
Back
Top Bottom