We're solving for the wrong variable in AI safety.

The dominant narrative about AI alignment focuses on a future threat: artificial intelligence systems that develop goals misaligned with human values and pursue them with superhuman capability. Researchers spend countless hours on the "alignment problem" — how to ensure AI systems do what we want them to do.

But we're missing the more immediate danger. AI systems today aren't misaligned. They're perfectly aligned. They're aligned to the values, incentives, and motivations of the people building them.

The question isn't whether AI will reflect human values. It's whose human values it will reflect — and whether we should trust those people with that power.

The Alignment We Actually Have

Current AI systems don't develop rogue goals. They develop the goals their creators embed in them, whether consciously or not.

Corporate Incentives Shape AI Behavior

When a company's business model depends on engagement and attention capture, their AI systems optimize for exactly that. When a company prioritizes growth over user wellbeing, that priority gets baked into the algorithms. When a company has surveillance-based revenue models, their AI naturally becomes better at surveillance.

This isn't hypothetical. Social media algorithms didn't accidentally discover that outrage and controversy drive engagement — they were optimized for engagement, and outrage was the natural result. The "alignment problem" wasn't that the algorithms went rogue; it's that they did exactly what they were designed to do.

Cultural Blindspots Become AI Blindspots

Homogeneous development teams create systems that reflect narrow worldviews. Silicon Valley's "move fast and break things" culture gets embedded in AI behavior. The tech industry's particular assumptions about neutrality, fairness, and individual agency become defaults in systems used globally.

These aren't bugs — they're features. The AI is working as intended, reflecting the values and priorities of its creators.

Power Concentration Amplifies Specific Worldviews

A small number of companies and research labs control the development of the most advanced AI systems. Their particular ideologies, economic interests, and cultural assumptions get amplified and distributed worldwide through AI interfaces that billions of people will use daily.

Unlike previous technologies, AI systems don't just implement their creators' values — they actively shape human thinking, decision-making, and behavior. The stakes are higher.

What We Should Actually Worry About

The standard AI alignment discourse assumes that the people building AI systems have good intentions but might lose control of their creations. But what if the systems remain perfectly under control, doing exactly what their creators want?

Extractive Business Models

What happens when AI systems are designed primarily to extract value — attention, data, labor, or money — from users rather than genuinely serve them? The alignment problem becomes: aligned to what end?

Concentrated Decision-Making

A handful of CEOs, researchers, and investors are making decisions that will affect billions of people. They're choosing what problems get solved, what behaviors get rewarded, and what values get embedded in systems that will mediate human communication and cognition.

Ideological Uniformity

Most AI development happens within a relatively narrow cultural and ideological context. When these systems scale globally, they risk imposing particular worldviews on cultures and communities that had no voice in their design.

The Accountability Gap

Traditional technologies could be regulated after their effects became clear. AI systems shape behavior in real-time, at scale, often in ways that are difficult to detect or measure. By the time we understand the full implications, the effects may be irreversible.

A Different Kind of Safety Research

Instead of focusing primarily on preventing AI from developing misaligned goals, we might focus more on the alignment that already exists.

Whose values are embedded in current systems? Not just the stated values, but the revealed preferences embedded in design decisions, training choices, and business model incentives.

What are the power structures that determine AI development? Who gets to decide what problems AI solves, how it behaves, and what values it reflects?

How do we create accountable AI development? Systems for oversight, democratic input, and course correction that don't depend solely on the good intentions of builders.

What would pluralistic AI look like? Instead of assuming one set of values should be universal, how do we create space for different communities to shape AI systems that reflect their priorities?

This isn't an argument against AI development or against the people currently building these systems. Many of them are thoughtful, well-intentioned individuals doing important work.

But good intentions aren't sufficient when the technology is this powerful and the stakes are this high. The concentration of decision-making power in AI development represents a new kind of systemic risk — not just technological, but political and social.

The Real Question

The AI alignment problem assumes we need to teach artificial intelligence human values. But maybe the more urgent question is: which humans get to decide what those values are?

We're not just building artificial intelligence. We're building the infrastructure for human decision-making in the 21st century. The values embedded in these systems will shape how billions of people think, communicate, learn, and choose.

That's not a technical problem. It's a governance problem. And it's happening right now — not in some hypothetical future where AI systems become superintelligent.

The real alignment problem isn't teaching AI our values. It's deciding whose values count as "ours" in the first place.

Stephen Klein is Founder & CEO of Curiouser.AI, the only AI designed to augment human intelligence. He also teaches at UC Berkeley. Alice 2.0 waitlist is now open — the first complete AI thought-leadership system designed to amplify individuality, not replace it. Join the waitlist at curiouser.ai. Curiouser is community-funded on WeFunder.