Felin and Holweg argue that AI and human cognition operate on fundamentally different principles, and that this distinction matters for how we think about innovation and decision-making. They push back against the assumption that large language models and other AI systems can replicate human thinking or replace human judgment. The core claim: AI works through data-based prediction—finding patterns in existing information—while humans use theory-based causal reasoning. Humans don't just process input and generate output like a computer. Instead, they build mental models about how the world works, then use those models to imagine possibilities that don't yet exist in any dataset. The authors highlight what they call "data-belief asymmetries" to show why this matters in practice. They use the example of heavier-than-air flight: the Wright brothers succeeded not because they had more flight data than their competitors, but because they had a different theory about how flight actually works. They could reason about aerodynamics causally, then design experiments to test their understanding. AI, by contrast, is "backward looking and imitative"—it can predict based on patterns in historical data, but it can't generate genuine novelty the way humans can by forming new causal theories and testing them through directed experimentation. The practical upshot is that human cognition has a mechanism for intervening in the world and creating new data through experimentation, whereas AI is stuck recombining and extrapolating from what already exists. This shapes everything from strategy to scientific discovery. The authors suggest that treating AI and human reasoning as equivalent misses something essential about how humans actually generate knowledge and make decisions under uncertainty, especially when the future looks nothing like the past.