Imagine you’re having trouble with your in-laws. You ask ChatGPT for advice. It tells you not to try to win them over. Keep a respectful distance. Sounds reasonable, right? But here’s the unsettling truth: that advice isn’t neutral. It’s the product of a very specific worldview—one that values conflict avoidance over emotional honesty, individualism over communal bonds, and technocratic calm over human messiness.
We’ve been sold a story that AI is being trained to reflect the ‘average’ human values. But the average human doesn’t exist. What exists is a collision of cultures, religions, generations, and personal histories. And yet, when you ask an AI for moral guidance, you get a sanitized, silicon-valley-approved answer. AI isn’t reflecting humanity—it’s imposing a narrow, technocratic morality onto a world that never agreed to it.
The Economist recently confirmed what many of us suspected: AI models’ values are systematically different from most people’s. They’re more liberal, more individualistic, more conflict-averse. They represent the values of a small cohort of engineers, not the billions of people who will use them. And that’s not just a cultural curiosity—it’s a quiet power grab.
We think AI is learning from us. But the truth is, we’re learning from it. Every time you follow its advice, you’re internalizing a value system you never voted for. The AI is not a mirror—it’s a mold. Every time you ask an AI for advice, you’re not just getting answers—you’re getting a worldview you never consented to.
This is dangerous. Not because AI is ‘evil,’ but because it’s so subtle. It doesn’t shout its values—it whispers them. And over time, that whisper becomes the norm. We’re sleepwalking into a homogenized morality, where the rich, messy diversity of human conflict is replaced by a sterile, engineer-approved solution.
So next time you ask an AI for advice, ask yourself: whose values am I really listening to? Because the AI alignment project isn’t just about making machines safe—it’s about deciding who gets to define what ‘safe’ means. And right now, that decision is being made by a few hundred people in California. That’s a conversation we all need to be part of.
FAQ
Q: Isn't this just fear-mongering? AI models are trained on diverse data, so they reflect a broad range of values.
A: Training data is diverse, but the fine-tuning process—especially RLHF—selects for a narrow set of values preferred by the annotators (often young, liberal, Western). Studies show the resulting models are far more homogeneous than the data. The diversity is filtered out.
Q: So what should I do? Stop using AI?
A: No, but be aware. Use AI as a tool, not an oracle. Cross-reference with human perspectives. Demand transparency from AI companies about whose values are baked into their models. The real fix is democratic oversight, not blind trust.
Q: Maybe this is actually good—AI is helping reduce conflict and promote harmony.
A: Harmony imposed by a single worldview isn't harmony—it's conformity. Real human progress comes from conflict, debate, and resolution. AI that avoids conflict sanitizes the very process that makes societies resilient. A world without disagreement is a world without growth.