They built a social network made only of bots. The result? Worse than humans.
In Amsterdam, researchers built a social network with no people at all. Inside were 500 bots posting, following, and reposting each other, all powered by ChatGPT. They wanted to answer a simple question: if we remove humans from social media, do hate and polarization disappear? You have no idea what happened: watch till the end, because what they found changes how we think about social media.
The answer is… no! Even without humans, the network creates bubbles, boosts extreme views, and focuses attention on a few accounts. The system itself produces the poison.
Petter Törnberg found that it’s not only toxic content that does the damage, it’s the network structure that pushes it up. Once an extreme post gains traction, a loop starts that keeps it on top and hides everything else. It’s not about “bad apples,” it’s the whole basket.
They tried to “fix” the platform. Six changes: chronological feeds, lower virality, hiding followers and reposts, removing bios, promoting opposite views. Nothing worked. Some changes made it worse. The chronological feed, for example, pushed extreme posts even higher. We see the same thing every day: the system rewards whoever shouts the loudest.
So the real problem is that with AI, things get serious. It’s not just human trolls anymore. Machines can now generate thousands of polarizing posts designed only to grab attention and make money.
The “digital town square” is a myth. It wasn’t killed by users, but by the system itself, now accelerated by AI.
We have our opinion. Tell us in the comments if you think social media can still be fixed or if we should start over. If you want to stay updated on how AI is rewriting the rules of social media, make sure you follow.
#ArtificialDecisions #MCC #AI
✅ This video is brought to you by: https://www.ethicsprofile.ai
👉 Important note: We’re planning the upcoming months.
If you’d like to request my presence as a speaker at your event, please contact my team at: [email protected]