Between You and AI

3Data Sensemaking

IMAGINE A 5,000-PIECE jigsaw puzzle of the beautiful Rio de Janeiro landscape, with the iconic Dois Irmãos mountain in the background. A stunning image—but a daunting challenge: As you sit before the scattered pieces, from where do you begin? What’s your instinctive approach to making sense of the chaos spread across the table?

Surely, you’ll rely on visual cues—colors, patterns, recognizable elements in the pieces. A blue piece might belong to the sky or sea, a face to another part of the image, and so on. It’s a slow, trial-and-error process, guided by intuition, memory, and a bit of luck. While this approach is creative, engaging, and even fun, we have to admit that it is also incredibly inefficient. You make mistakes, misplace pieces, and spend lots of time searching for just the right fit, constrained by your cognitive limitations.

Now, consider how an AI system would tackle the same puzzle. Instead of relying on colors, visual patterns, or familiar imagery like humans do, the AI approaches the problem from a completely different angle—actually based on the edges and not on the content of the pieces. Rather than getting lost in the complexity of the image, it focuses on structural and mathematical patterns that are almost invisible to us.

For instance, instead of seeing a yellow piece and thinking, “This must belong to the sun,” AI categorizes the piece purely by geometric properties—edge angles, connector depth, and statistical fit within the overall ...