Social media ‘echo chambers’ amplify antisemitism

Social media algorithms are leaving Jews “defenceless” against antisemitism, and right now, there’s probably the best distribution methodology for spreading antisemitic ideology in history, said Steven Sidley, professor of practice at Johannesburg Business School and expert in artificial intelligence (AI).

“Previously antisemitism arose in countries – the most obvious being Nazi Germany – where they took control of the entire media-distribution platform and held German citizens captive with antisemitic messages. Through this practice, the Nazi regime was able to get people to believe that the Jews were vermin,” Sidley said.

However, today with social media, anyone has the ability to reach billions of people with incredibly effective negative messaging, Sidley said.

“Take for example short TikTok videos, none of them have to be factual, and there’s no authority checking the factuality of antisemitic content,” he said. “Some companies, such as Facebook, used to have internal fact-checking departments to put a lid on the amount of misinformation sprouting. However, under Trump, they have stepped back, rid their companies of fact-checking departments, and claimed free speech on an open platform.”

Sidley said there was absolutely no filter anymore, and AI has supercharged the algorithms. “AI not only can spot what you like easily, but it also catches you just by analysing the time you spend on a certain post,” he said. With all this information and analysis, AI can decide whether a consumer will be vulnerable to certain messaging and misinformation. “The targeting of messaging has become incredibly powerful,” Sidley said.

Benjamin Rosman, professor of AI and robotics at the University of the Witwatersrand, agreed, saying, “The techniques used in social media are recommendation systems. Their internal objective is to maximise engagement.

“Social media companies figured out they can draw users in to spend more time on their platforms by changing the content the user is seeing,” Rosman said. In the early days of Facebook, users were viewing only what their friends posted. By implementing algorithms, the recommendation systems, the social media companies can choose what content to show the user to control how much time the user spends on their platform, Rosman said.

The algorithms aren’t trying to learn what the user likes, rather they focus on figuring out what will capture the user to keep them engaged longer, Rosman said. He said the recommendation systems are looking at the time a user spends looking at a particular post or if the user clicks on certain posts. The system then finds similar content to post on the user’s feed based on how the user has engaged with previous content.

Sidley said algorithms monitor where the consumer is spending their time, and where their attention is focused. In turn, the algorithm will then amplify other content that’s likely to keep the consumer engaged. “There are two big attention grabbers on social media at the moment, the amplification of outrage, and content that’s funny,” he said.

Social media is an attention grabbing and addictive mechanism, Sidley said. Various studies have determined that it works on dopamine. When a consumer views content catered to them, not only do they feel a kick of neurochemicals responding as their outrage or amusement peaks, they get to feel part of a community along with the others that share their interests in content,” Sidley said.

Rosman said, “Users are playing to their confirmation biases. Users get outraged and infuriated, and share the content they’re viewing with family groups and friends. The system’s perspective is that the user clearly engages with this type of content and will in turn bombard the user’s feed with even more outrageous content because the human brain tends to react strongly to content that outrages a person.” Rosman said that what ends up happening is that like-minded people start creating spiralling groups together, playing into users’ confirmation biases. If a user thinks there’s a problem with the world, the platform will keep reinforcing it to hold the user’s attention.

“The problem is, the more outrageous the content, the more it slips into the realm of conspiracy theories. Often the further one deep dives into conspiracy theories, they tend to include some sort of antisemitic tropes,” said Rosman.

Sidley agreed, saying, “We’re in a situation where non-antisemites, particularly Jews, are quite defenceless against this, more so than ever.

“The only thing one could theoretically do is to have diametrically opposed messaging go out. But Jews are a small population with limited resources. They can’t possibly put out as much content as the other side is producing. Even if [Jews] are putting a counter position out, if the consumer isn’t liking any of their content, the algorithm will no longer expose them to it. The algorithm builds the echo chamber because it knows what keeps your attention.

“The algorithm homes in on what makes a person tick to keep them engaged. The only way to change the content one is viewing is by overriding one’s primal instincts of anger. The user must consciously redirect their attention if they want to be shown different content,” said Rosman.

Rosman cites an example. “If a user engages with a pro-Israel post, the algorithm won’t show the user any anti-Israel content because it would rather feed into the pro-Israel sentiment as it has learnt that this type of content engages the user.

“It won’t show both sides because it isn’t designed to be nuanced. It’s designed to give the user more intense versions of a previous post that caught their attention. The algorithm tries to anticipate what would hold one’s focus to keep them on the platform for as long as possible by showing closely related content.”

Said Rosman, “More time spent means more money for the social media company. Their incentive is to keep you engaged rather than the user engaging with nuanced and unbiased conversations because nuanced conversations don’t tend to support their goal.”

Neither Rosman nor Sidley have any clear plan for how to turn this dangerous freight train around. They suggest that Jews actively start mass producing their own content, and that users wake up to the biased, misinformed content on their feeds, and seek out facts.