The Montreal Economic Institute published a striking piece this week arguing that no matter how powerful AI becomes, it can't replace market-based economies. The core claim: The economy isn't a math problem computers can solve. It's a constantly evolving discovery process happening in real time, driven by billions of people making choices based on information only they possess. As AI capabilities explode—with models now solving 72% of real-world coding problems compared to just 4% in 2023—some have started wondering whether super-intelligent machines could finally make central planning work. The answer, according to the economic research, remains a definite no.

This connects directly to work from the American Enterprise Institute, where economist Lynne Kiesling explains that central planning fails because it treats economic calculation as a computation problem when it's actually a cognitive problem of diffuse private knowledge. The issue isn't computing power—it's that most human knowledge can't be captured in data at all. Knowledge is personal, often tacit, and frequently something we don't even realize we know until we act on it. No one knows your preference for milk or what you're willing to pay until you actually buy a quart for $3.99—only then does that private knowledge become visible data. This distinction between knowledge and data turns out to be critical, and it's why even the most advanced AI can't substitute for markets.

Here's how it works: Markets transform knowledge into information through the price system, but that transformation only happens when people act on their private knowledge through buying and selling. Think about choosing a restaurant tonight. You know you're tired, that you had Italian yesterday, that your budget is tight this week, and that the place on the corner has slow service. None of that knowledge exists as data anywhere until you make a choice. Some assume that because computation has become cheap and AI systems can handle large-scale optimizations, the planning problem is solved—but that falsely identifies economic calculation as a computation problem. AI can crunch numbers brilliantly. What it can't do is know what numbers to crunch before people reveal their preferences through actual market choices. Markets play the role of knowledge ecosystems coordinating across billions of strangers so each can achieve their plans.

This isn't just theoretical hairsplitting. It explains why companies like Amazon succeed with internal planning while socialist economies failed—Amazon operates within a broader market that generates the price signals making rational calculation possible. The distinction between knowledge and data bounds us strictly away from visions of replicating market outcomes in a centrally-planned system, even with advanced AI. As AI reshapes work and productivity, it'll be a powerful tool within market economies. But replacing those markets entirely? The knowledge problem says that's impossible, no matter how smart the machines get. The economy isn't waiting to be optimized by a sufficiently large computer. It's an ongoing conversation among billions of people, conducted in the language of prices, discovering what's valuable one transaction at a time.