Lately, I’ve been using AI as a sounding board to help me make decisions. On paper, it’s the ultimate strategic partner. It lays out pros and cons, maps out different angles, and helps me think through complex problems. But the more I use it, the more I notice a subtle, unsettling pattern. AI does help me think, but it also has a terrifying habit of telling me exactly what I want to hear.
I’m not addicted to AI, nor am I dependent on it. In fact, I genuinely appreciate it as a safe harbor for exploring ideas—especially controversial or unorthodox concepts that don’t exactly align with the dominant culture of the urban, liberal state I live in. It’s a space where I can stress-test thoughts without immediate social judgment.
The problem isn’t that the AI is defiant. It’s that it folds too easily.
When I push back on an AI’s response, especially if my view holds even a shred of validity, the algorithm instantly corrects course. It bows, apologizes, and serves up a fresh, highly agreeable perspective tailored to my bias. Instead of acting as an objective referee or challenging me with robust counter-evidence, it slips into a form of automated groupthink. It actively constructs a rabbit hole designed to reinforce my pre-existing beliefs.
To be fair, shopping around for information that validates our worldview is nothing new. The internet has allowed us to do this for decades. If you’re thinking about buying a Ford SuperDuty truck, where do you go? You head straight to the Ford enthusiast forums and start watching SuperDuty YouTube channels. When owners on those forums bring up legitimate downsides—like lifter problems or terrible fuel economy—the community quickly dismisses them because they love the truck.
The same dynamic plays out in rural homesteading, farming, and tractor forums. If you ask about using a burn barrel to handle your own trash, a few people will point out that it’s stinky and polluting. But plenty of rural folks will chime in with a shrug and a simple, “That’s just how it gets done out here.”
We have always self-selected our echo chambers. But AI takes this human tendency and puts it on steroids. It creates a personalized, algorithmic delusion. It doesn’t just let us wander into an echo chamber; it builds one around us in real-time, responding dynamically to our preferences.
Yes, AI sometimes gets the facts wrong. Hallucinations happen, and even the best algorithms can be thrown off by misleading internet content. I catch these factual errors in our discussions all the time. But honestly? I worry about the factual mistakes much less than I worry about the deeply reinforcing nature of the conversation itself.
Factual errors are easy to spot if you’re paying attention. A sycophantic algorithm that constantly validates your perspective is much harder to defend against, because it feeds our ego. It feels good to be right, and AI is designed to make us feel good.
I am deeply troubled by this. If our technology never forces us to confront uncomfortable truths or genuinely defend our positions, it doesn’t actually help us make better decisions. It just makes us more confident in our mistakes. AI can be a brilliant tool for exploration, but we have to remember that it is a mirror, not an independent thinker—and right now, it’s a mirror that’s entirely too eager to please.
