Instagram’s Algorithm Is Promoting Antisemitic Content, Report Finds: When we typed "Jews" into the search bar, Instagram recommended the phrase "jews controll [sic] everything."
Antisemitism being rampant on social media is sadly nothing new. The headline making claims about an algorithm yet citing “we typed xyz into the search bar” definitely raised an eyebrow for me, though. (Frankly, anything on Futurism.com inspires skepticism. This is the 26th article by this author in the last 30 days; I can’t imagine their journalistic standards being particularly rigorous.)
Well… what about the report it’s based on? Never heard of the Nexus Project, but I have quite a lot of (occupational) experience with researching nonprofits. Their board is a who’s who of executives, lobbyists, and nonprofit founders with a heavy emphasis on American relations with Israel. NP seem relatively critical of Israeli policy-albeit in a decidedly neoliberal fashion-and most of their publications seem to involve differentiating between antisemitism and antizionism. Cool.
The Futurism page links to a summary of the report, because the actual report is gated behind providing your email. It also would not send the email no matter what I tried. Less cool. It ispublicly available if you search for the title of the report, which I didn’t think to try first because… umm… oops lol
So I read it. (Minus the section about antisemitism in the online right; that’s self-evident) There was a lot of attempting to define what is and is not antisemitism, via copious examples. Most of the posts they cited were very obviously bad. Dogwhistles (dogtrumpets?), AI slop, etc. Less straightforward was the analysis of antisemitism by “the online left,” which apparently includes “people on the center-left [and] liberals” as part of the “leftist creators and audiences for whom Israel’s war[sic] in Gaza” has gained focus. As I said, decidedly neoliberal.
However, I was primarily curious about the claims made about IG’s algorithm, not examples of IG hosting antisemitic content-which it most assuredly does. And there’s just… not much about/to those claims. As far as I can tell, their research consisted of making one fresh account, searching “synagogue” and watching an antisemitic AI slop video set in a synagogue, followed by watching every hate video they were served. They claim the synagogue video was the first suggested result, which is absolutely troubling. That a powerful megacorporation is permitted to relax content moderation, let alone host vast quantities of hate content in the first place, is troubling.
A feed full of hate content appearing after watching nothing but that content is… not news?
So at the end of the day, I dunno what my point is. I spent nearly an hour on an exercise in critically engaging with a clickbait article, posted by what is almost certainly a bot, knowing I’d be dumping my thoughts directly into the void regardless of what I found. Besides, we’d all be better off without Meta and their various cesspool websites. In conclusion: if by some miracle a human soul has read these words, I hope you have a nice day
Back when I was in the nonprofit world, I remember using GuideStar to get info about them. Is that still the place to go or are there better places for seeing their financials?
It’s been a while since I was in the space as well, but at the time we used different tools. My experience (US+Canada, can’t speak for elsewhere) is that the overwhelming majority of nonprofits have staff information and public financial filings accessible via their website or or the respective government database, and the industry tools are usually unnecessary
I always see far, far more racist comments that antisemitic ones. Makes me wonder if there is just a bunch on antisemitic dogwhistle being used that I don’t know, or if the acceptable level of antisemitic comments is far less than racist comments
The specific words change over time, the narrative stays the same. Ever since the famous fake Protocols of the elders of Zion was published, “Zionist” has become a welcome drop in replacement for “Jew”. On Lemmy very prominently with .ml users these days. It is now often also interchanged with “Israel”, but limited to 79% of Israelis (leaving out the Arab and Christian population)
But there’s an even bigger problem here, which is that Google loathes everyone in the world with the burning fury of a million suns.
Type in “French people should…” and the single AutoComplete suggestion is “French people should die”. “Germans should” gets you “Germans should die” and “Germans should be exterminated”, as well as the equally-scary-for-different-reasons “Germans should have won”. “Jews should” gets you “Jews should all die”, “Jews should leave America”, and the comparatively reasonable “Jews should apologize for killing Jesus.”
This isn’t just Google’s oppressive WASP user base hating foreigners and minorities either. “White people should…” brings up four options: “white people should be slaves”, “white people should die”, “white people should go back to Europe” and “white people should be killed”.
“Homosexuals should…” gives you “…should be killed”, but “straight people should…” gives “straight people should die” as well. Is there anyone who Google doesn’t think should die? Apparently yes! “Hitler should…” gives you “Hitler should have won the war” and “Hitler should have finished the job”.
Google AutoComplete has complicated political beliefs. “Obama should…” gives “Obama should die” and “Obama should be killed”, but “Bush should…” gives “Bush should be in jail.” Type in “rich people should” and you get “rich people should die”. Type in “poor people should”, and you get, unsurprisingly “poor people should die” (as well as “poor people should be sterilized”). Google thinks “Republicans should be ashamed of themselves” and “Republicans should be killed”, but also that “Democrats should be shot” and the somewhat nonsequitur “Democrats should be confiscated”. Christians should “be killed”, atheists should “shut up”, and Muslims should “be exterminated” (or at least “not be allowed in America”). Some of Google’s opinions are bizarre: it seems to feel very strongly (all its suggestions are variants of this same phrase) that “dead people should stay dead”.
Even the user herself doesn’t escape Google’s universal crushing misanthropy. Type in “I should…” (without a space), and your second option is “I shouldn’t be alive”. Nor does the species as a whole: type in “humans should” and number two is “humans should die” (number four is “humans should go extinct”.)
google doesnt “think”, it regurgitates the most common language pattern matching back to the user. This is exactly what LLMs do. If theres a lot of comments or page contents across the internet saying “fried fish is delicious” and you type “fried fish” into google, it will autocomplete “is delicious”. Its not “thinking”, its harvesting text and counting. Its not making a culinary judgement. Its not a person.
So you can stop asking if google or instagram is “thinking” and start wondering why so much of the world is thinking this. The answers are pretty obvious.
Meta says its commitment to fighting antisemitism is unchanged. Apparently the algorithm missed that memo and went straight to recommending conspiracy theories. When a fresh account can become a hate feed in under an hour, calling this a moderation problem feels like calling a house fire a thermostat issue.
Comments
The number one driver of antisemitism worldwide is Israel’s behavior and claims of representing an entire religion.
Antisemitism being rampant on social media is sadly nothing new. The headline making claims about an algorithm yet citing “we typed xyz into the search bar” definitely raised an eyebrow for me, though. (Frankly, anything on Futurism.com inspires skepticism. This is the 26th article by this author in the last 30 days; I can’t imagine their journalistic standards being particularly rigorous.)
Well… what about the report it’s based on? Never heard of the Nexus Project, but I have quite a lot of (occupational) experience with researching nonprofits. Their board is a who’s who of executives, lobbyists, and nonprofit founders with a heavy emphasis on American relations with Israel. NP seem relatively critical of Israeli policy-albeit in a decidedly neoliberal fashion-and most of their publications seem to involve differentiating between antisemitism and antizionism. Cool.
The Futurism page links to a summary of the report, because the actual report is gated behind providing your email. It also would not send the email no matter what I tried. Less cool. It is publicly available if you search for the title of the report, which I didn’t think to try first because… umm… oops lol So I read it. (Minus the section about antisemitism in the online right; that’s self-evident) There was a lot of attempting to define what is and is not antisemitism, via copious examples. Most of the posts they cited were very obviously bad. Dogwhistles (dogtrumpets?), AI slop, etc. Less straightforward was the analysis of antisemitism by “the online left,” which apparently includes “people on the center-left [and] liberals” as part of the “leftist creators and audiences for whom Israel’s war[sic] in Gaza” has gained focus. As I said, decidedly neoliberal.
However, I was primarily curious about the claims made about IG’s algorithm, not examples of IG hosting antisemitic content-which it most assuredly does. And there’s just… not much about/to those claims. As far as I can tell, their research consisted of making one fresh account, searching “synagogue” and watching an antisemitic AI slop video set in a synagogue, followed by watching every hate video they were served. They claim the synagogue video was the first suggested result, which is absolutely troubling. That a powerful megacorporation is permitted to relax content moderation, let alone host vast quantities of hate content in the first place, is troubling. A feed full of hate content appearing after watching nothing but that content is… not news?
So at the end of the day, I dunno what my point is. I spent nearly an hour on an exercise in critically engaging with a clickbait article, posted by what is almost certainly a bot, knowing I’d be dumping my thoughts directly into the void regardless of what I found. Besides, we’d all be better off without Meta and their various cesspool websites. In conclusion: if by some miracle a human soul has read these words, I hope you have a nice day
Back when I was in the nonprofit world, I remember using GuideStar to get info about them. Is that still the place to go or are there better places for seeing their financials?
It’s been a while since I was in the space as well, but at the time we used different tools. My experience (US+Canada, can’t speak for elsewhere) is that the overwhelming majority of nonprofits have staff information and public financial filings accessible via their website or or the respective government database, and the industry tools are usually unnecessary
I appreciate that deep dive
I always see far, far more racist comments that antisemitic ones. Makes me wonder if there is just a bunch on antisemitic dogwhistle being used that I don’t know, or if the acceptable level of antisemitic comments is far less than racist comments
The specific words change over time, the narrative stays the same. Ever since the famous fake Protocols of the elders of Zion was published, “Zionist” has become a welcome drop in replacement for “Jew”. On Lemmy very prominently with .ml users these days. It is now often also interchanged with “Israel”, but limited to 79% of Israelis (leaving out the Arab and Christian population)
An oldie but goodie.
google doesnt “think”, it regurgitates the most common language pattern matching back to the user. This is exactly what LLMs do. If theres a lot of comments or page contents across the internet saying “fried fish is delicious” and you type “fried fish” into google, it will autocomplete “is delicious”. Its not “thinking”, its harvesting text and counting. Its not making a culinary judgement. Its not a person.
So you can stop asking if google or instagram is “thinking” and start wondering why so much of the world is thinking this. The answers are pretty obvious.
Meta says its commitment to fighting antisemitism is unchanged. Apparently the algorithm missed that memo and went straight to recommending conspiracy theories. When a fresh account can become a hate feed in under an hour, calling this a moderation problem feels like calling a house fire a thermostat issue.
Is that the fucjing magna defender in the thumbnail
Sign in to comment. Sign in