Redazione RHC : 20 September 2025 09:36
Researchers at the University of Amsterdam conducted a novel experiment, creating a social network populated exclusively by artificial intelligence bots. The primary objective was to analyze how digital dynamics could be influenced by different moderation measures.
The underlying idea was to test the effectiveness of tools capable of reducing two of the most serious problems of contemporary social media : polarization and toxicity in online interactions.
To this end, six different strategies were tested. These included switching to a chronological feed, increasing the visibility of dissenting opinions, hiding subscriber statistics, and removing profile descriptions.
For example, sorting the news feed chronologically reduced attention inequality, but meant that more extreme content was pushed to the forefront. This is a sobering reality, and one that contrasts with companies’ promises to build a “digital town square,” as
Elon Musk called it) where everyone can coexist peacefully.
Despite expectations, the results have been less than encouraging. None of the tested solutions actually reduced polarization within the simulated platform.
On the contrary, some measures even made the situation worse. The chronological feed, for example, did reduce inequality in user attention, but it also favored the rise of extreme content, bringing it to the top of visibility.
Scholars have emphasized that toxic content tends to create resilient networks, capable of sustaining and amplifying its spread, regardless of attempts to limit its impact.
With the advancement of generative artificial intelligence, the amount of this content is expected to grow further. This represents a profound challenge to traditional social media models and makes the search for radical changes to ensure healthy digital communities ever more urgent.
Redazione