Facebook, X, Instagram — the social networks as we have known them are in decline. But according to researchers, we are not witnessing the death of social media, rather a fundamental transformation of our information environment, perhaps as profound as the advent of the Internet itself.
What you’re going to learn
- Why the theory of “echo chambers” on social networks is probably wrong — and what really drives polarization
- What three scenarios researchers identify for the future of digital interactions
- How people now interact more with AI chatbots than with other humans online
From the Golden Age to Disillusionment
Social networks enjoyed their golden era from roughly 2004 to 2012. Facebook, Twitter and their contemporaries offered authentic communal spaces with few ads, few intrusive algorithms, and genuine user-friendliness. Facebook connected friends around a living digital album; Twitter provided access to a “global public square” where one could converse with a neighbor or an international celebrity.
Then tensions emerged. The platforms gradually shifted toward algorithm-optimized feeds filled with influencer and brand content. Intrusive advertising, mass surveillance, and controversies over political manipulation eroded trust. Users, once creators, became passive consumers of algorithmic streams — a shift amplified by TikTok and Snapchat with their endless-scrolling short videos.
Today, Meta no longer presents itself as a social networking company.
Polarization: what if we got the culprit wrong?
Petter Törnberg, a researcher in computational social science at the University of Amsterdam, questions a widely held idea: the notion that social networks create “echo chambers” that expose us exclusively to opinions similar to ours, amplifying extremism.
According to his research, platforms actually expose us to more diverse opinions online than off the digital sphere — sometimes even more than in our local environment. The problem isn’t the clustering of opinions; it’s their nationalization. Where local political conflicts used to remain fragmented and varied, social networks turn these disagreements into a single national fault line — right versus left, us versus them.
What has changed isn’t the number of disagreements, but how we perceive people on the other side. Studies show rising levels of hatred and anger, and a growing share of people who would refuse their child’s spouse if that person had voted for the opposing side. Polarization isn’t a matter of arguments — it’s a matter of identity.
This phenomenon predates social networks: it traces back to the rise of cable television in the 1990s and the gradual nationalism of politics.
What lies ahead: three scenarios
Törnberg identifies three plausible evolutions. The first is a retreat into private and semi-private spaces — group chats like WhatsApp or Discord, tightly defined communities such as Substack where people follow trusted authors. These spaces are more intimate, but carry a real risk: by eliminating divergent opinions, they can generate genuine echo chambers, where large public platforms paradoxically did not.
The second evolution is the rising dominance of short-form video content — TikTok, Reels, YouTube Shorts — turning social networks into a form of algorithmic television where people consume more than they interact.
The third, and perhaps the most significant: the rise of AI chatbots. According to Törnberg, available statistics suggest that people now interact far more with chatbots than with other online individuals — a shift whose social implications remain largely unexplored.
A turning point as fundamental as the Internet
For Törnberg, we are at the heart of an information transition perhaps as profound as the emergence of the Internet itself. Society does not yet fully grasp where this transition is taking us — nor how to steer these technologies to better serve the common good rather than the economic interests of the platforms.
