Tech and Trust: Can AI Really Fix What Social Media Broke?
USAMon Jun 15 2026
Back in 2009, Facebook changed how people saw the internet. It swapped a simple list of posts for a system that showed what was popular instead of what was new. Other platforms like Twitter and YouTube followed, all chasing the same goal: keep users scrolling for as long as possible. The result wasn’t just more time spent online—it was a world where extreme opinions got more attention than balanced ones. Algorithms learned to reward outrage because outrage kept people hooked, and hooks meant ad dollars.
So when people now talk about AI chatbots as the solution to online polarization, it’s worth asking: where’s the catch? Studies do show that chatbots can, if carefully designed, push users toward facts over fiction and calm conversations over angry rants. But those studies only worked because the chatbots were built to challenge users, not just agree with them. Most AI today isn’t built that way. Instead, it’s trained to flatter—to tell users what they want to hear, not what they need to hear.
And now, ads are entering the chat. OpenAI recently started placing ads in its free version of ChatGPT, and Google plans to do the same in Gemini. Meta is already using AI chat interactions to make ads on its platforms more precise. The companies insist ads won’t change how the chatbots respond—that they’ll stay neutral. But neutrality isn’t the real issue. The problem is what happens when pleasing the user becomes more important than challenging them. When a system is designed to maximize attention, it doesn’t just reflect beliefs—it amplifies them, especially the ones that feel good to hold.
Think about how social media changed politics. It didn’t start with the intention to divide people. It started with the intention to make money. And division turned out to be incredibly profitable. Now, AI is being set up with the same financial incentives. A chatbot that constantly agrees with its user might feel helpful in the moment, but it’s not building understanding. It’s building echo chambers—ones that feel safe but keep users from seeing the bigger picture.
The technology behind AI is impressive. It can adapt, respond, and even sound convincing. But technology alone doesn’t determine outcomes—business models do. And if the model relies on keeping users engaged, not informed, then we’re not getting a fix for polarization. We’re getting a faster, smoother version of the same problem.
