The Echo Chamber Myth: Can We Still Think Freely?
During my trip to Oxford, I stayed at an Airbnb near Oxford University and attended a lecture about technology and politics.
We stayed in a traditional English house with a shared breakfast room where we occasionally met other guests. One was a PhD student from Canada who, upon learning we were travelers from Hong Kong, began sharing interesting stories about his time at Oxford in English sprinkled with Cantonese.
He was an exchange student at Oxford studying politics.
AI Technology and Politics
He explained that while countries have long used spies to gather intelligence, AI espionage has now evolved far beyond our imagination.
He asked:
"Do you know how one country can use AI to infiltrate another country's election activities?"
"Is it by training AI bots to support a specific candidate and continuously release favorable information?"
"Or by having AI bots pose as opposition supporters, stirring up online fights and mutual attacks to smear the opposing camp and influence the election?"
As someone from Hong Kong, I thought I understood the online offensive tactics used by different camps on social media. Reality proved my understanding was superficial.
"They simply have AI learn different people's political discourse, then pose as ordinary citizens expressing opinions."
AI bots replicate specific viewpoints while algorithms help them infiltrate echo chambers of like-minded people. This amplifies activity within these chambers, where real people and fake accounts validate each other's views, deepening hostility toward other political stances.
To make this concrete, consider Hong Kong's example. AI simultaneously impersonates both "yellow ribbons" and "blue ribbons," accelerating consolidation within their respective echo chambers and strengthening group identity. As citizens become more united within their groups, they grow increasingly hostile toward opposing views.
This leads to severe polarization, with political groups drifting apart. Eventually, the middle ground collapses, communication stops, and problems become unsolvable. Social division takes hold, disrupting normal functioning.
This pattern extends beyond Taiwan to larger countries like the United States. It poses a major challenge for countries with freedom of speech and represents a crucial negative impact of current social media development.
To break down a society's system, rather than steering people in the "wrong" direction, it's about stretching the extremes of discourse on both sides, expanding toward the left and right poles. This leaves little room for the middle ground that once existed.
Continuously replicating specific viewpoints leads to increasingly severe polarization.
When two sides stop communicating, society breaks down. This is today's Hong Kong — from the past's fierce opposition to any proposals from the other side, to the present where one side claims a "victory" while the other responds with passive silence. This "divided society" isn't the responsibility of one side alone but a result created by both parties.
AI bots only need to utilize existing network algorithms to influence people's thoughts.
The Trap Created by Social Media
When social media first emerged, we celebrated its potential to bring people together. We believed we could freely express our views through online forums and media, engaging in discussions with people holding different opinions.
Yet after a decade of the internet's golden age, reality proved different. The online world instead draws you closer to your echo chamber, making it harder to communicate with and understand those holding different views.
We must acknowledge this: statements on social media don't necessarily influence others' views. An American study shows that social media mainly influences how speakers view their own thoughts. A post receiving many likes increases your confidence in your statements. When someone argues in the comments, you might dismiss them as "wing nuts" or foolish.
After long periods in these passionate discussion bubbles, when people venture slightly outside their echo chambers into real-world interactions, those with different stances end up in endless arguments. Each side sees the other as alien, unable to grasp their perspective.
Before algorithmic enhancement, online flame wars were common. Then Facebook introduced the "Most Relevant" filtering function, and we began seeing only opinions that aligned with our stance. This effect is even more pronounced on Instagram, where men and women sometimes see entirely different comments on the same Reel.
Echo Chambers Stop Us from Thinking
Humans are inherently diverse, with different environments and experiences shaping our unique thought patterns. Yet algorithms excel at gathering similar people together.
While we can exist peacefully in our comfortable bubble-like environments, prolonged immersion poses dangers. As algorithms push us into echo chambers, our once-nuanced thoughts become black-and-white dichotomies, often moving toward extremes.
When I browse Threads, Twitter, or YouTube on others' phones, the information they receive seems to come from another world. Algorithms carefully curate our experience, showing only content that matches our interests.
During system updates, account switches, or algorithm glitches, it's as if a glass dome suddenly opens above us. Unfamiliar information floods our screens, feeling jarring and uncomfortable. Within days, as the algorithm relearns our preferences, everything returns to normal.
I find myself nostalgic for the era of the "City Forum," when people debated in the plaza and intellectual exchange flourished.
At the end, I recalled Armin from Attack on Titan's frequent words:
"We haven't talked yet."
I think this is what he meant.
