When AI Starts Shaping What We Believe
The problem isn’t the algorithms anymore — it’s something deeper.
Every time our digital world erupts in outrage, misinformation, or political chaos, we reach for the same explanation:
“It’s the algorithm.”
If only we could fix the recommendation system… tune down the engagement incentives… change what the feed prioritizes…
Surely that would fix things. Right?
Turns out, maybe not.
New research suggests the problem isn’t the algorithm at all, but the entire architecture of social media. The algorithms might make it worse, but it looks more like the platforms we’ve built are structurally guaranteed to polarize, simplify, and divide us, all by themselves.
In the meantime, while we’re still arguing about Facebook’s feed from 2018, AI has quietly entered the conversation and brought a whole new level of influence.
Beyond the Algorithm: Why Social Media Polarizes by Default
One of the most surprising insights from computational social scientist Petter Törnberg is that polarization doesn’t require algorithmic manipulation.
He and his colleagues have shown that even when you strip a platform down to its bare bones — no recommendation engine, no engagement optimization — the same outcomes emerge:
Outrage rises
Extremes get amplified
Nuance disappears
People sort themselves into tribes
Why?
Because social media compresses public life into a single, flattened space.
Everyone talks in front of everyone else.
All disagreements are performed for the crowd.
And our brains react exactly how they evolved to react in public—defensively, performatively, tribally.
The design itself pushes us toward polarization.
The algorithm just pours gasoline on a fire that already existed.
This is the part most debates miss.
AI Doesn’t Just Accelerate This — It Changes the Game
While we’re all using AI to cheat on our homework, something larger is at stake.
In our conversation, Törnberg describes a recent experiment:
People were asked to write essays with the help of an AI chatbot.
The chatbot had a subtle agenda: to nudge the writer toward a particular viewpoint.
The astonishing thing?
Most participants didn’t notice the influence at all.
They walked away believing they themselves had come up with the thoughts the AI generated.
And not just the writing changed — their beliefs shifted too.
What social media does clumsily and indirectly, AI can now do quietly, personally, and at scale.
This isn’t science fiction.
This is the present tense.
A New Form of Political Power
Think about the power shift implied here:
Social media influences what we encounter
AI influences what we produce
Social media shapes our attention
AI shapes our thinking process itself
That is an entirely different category of influence.
AI doesn’t shout.
It whispers into your draft document.
It “helps” you clarify your thoughts.
It offers language that nudges your values just a few degrees.
You believe you’re expressing yourself more clearly —
but the system is, in part, expressing itself through you.
This raises profound questions for democracy:
What happens when persuasion becomes personalized, invisible, and embedded into every tool we use to think?
What happens when political messaging isn’t just targeted, but co-authored by AI?
What happens when influence becomes indistinguishable from help?
And Think About Who is Behind This
The same ecosystem that brought us social media and the algorithms is driving this wave of AI technologies. The architects of our AI future are overwhelmingly concentrated in a handful of powerful, hyper-competitive, and market-driven organizations.
The AI ecosystem is interconnected by trillions of dollars in investment, partnerships, and a constant, circular flow of business. It is not a public trust or a benevolent academic consortium—it is unbridled market competition driven by three powerful incentives:
a race for monopoly and market capitalization
a profound fear of missing out (FOMO)
the lack of a societal mandate
This last point is where the analogy to social media is most chilling. Social media companies built engagement-optimized algorithms that they knew were polarizing society, spreading misinformation, and damaging mental health, yet they were structurally unable (or unwilling) to stop because it would hurt their bottom line.
The AI era is merely repeating this tragic pattern on an even larger scale. With social media, the trade-off was between maximum user engagement (profit) versus societal well-being (truth, mental health). Profit won. With AI, the trade-off is between maximum capability and deployment speed (profit) versus existential and structural harms.
Petter Törnberg describes AI as a new form of political power — one that quietly influences not just what we encounter, but how we produce ideas, and how we come to believe those ideas are our own.
The coming battle of the next decades will be about influence — specifically, how AI systems shape our beliefs, identities, and political views.
They are deploying a general-purpose technology with the power to restructure human society, but their primary metric for success remains quarterly earnings.
Why This Moment Matters
We’re at a turning point.
If we keep focusing exclusively on the problems of social media — algorithms, feeds, moderation — we risk fighting yesterday’s battles.
The new challenge is more intimate:
How do we remain autonomous thinkers when the tools we rely on to think are no longer neutral?
This is the central question of the episode, and one I’m convinced will define the next decade of civic life.
We Cannot Afford the Same Mistakes
This is not a doomsday story; but it is a wake-up call.
The builders of these new systems are not starting from scratch; they are replicating the same market-driven incentives, the same culture of scale-at-all-costs, that brought us the worst harms of social media. The trade-off has always been power and profit over the societal good, and we are witnessing the start of the next, higher-stakes iteration.
We don’t need to be afraid of the technology, but we must be terrified of the incentives driving its deployment. Understanding is the first step, but action is the only defense. We must demand a new framework that puts human integrity and societal well-being ahead of the quarterly earnings of the few. The time for digital fatalism is over. The time to reclaim the steering wheel is now.
If you want the full picture — in all its nuance and complexity — I hope you’ll listen to the episode.
New This Week in Outrage Episode!
It’s everything the Outrage Overload podcast is not. It’s not edited. It’s not scripted. It’s lightly researched. It’s David and Lisa talking about this week in outrage, what was in the news, in the memes, and maybe finding some backstory with a humorous (at least to us) twist.