The researchers had measured ten dimensions for each pilot, so now Daniels set himself the task of going through the data, pilot by pilot by pilot, and counting how many of the 4,063 pilots were in the middle 30 percent on all of the ten dimensions.

The answer, when it came, was this: none. There were no average-sized pilots—none whatsoever. Even if you looked at just three of the ten dimensions, fewer than 5 percent of the pilots were average-sized on all three. Even in a population of humans deliberately selected against a set of criteria (if you were too tall or too short, for instance, you weren’t qualified to become a USAF pilot in the first place), there was no one-size-fits-all, not even close.

Just as Don Clifton discovered that the only predictor of performance was total score across a number of relevant variables—that there was no right pattern of abilities, only a right sum of abilities—Gilbert Daniels discovered that there were no average humans in a population of 4,063, and that the average is a mathematical concept, not something that exists in the physical world. While the outcomes of high performance are visible and clear, the ingredients of high performance vary from person to person. There is no one-size-fits-all when it comes to human beings; and 8 9 there is no one-size-fits-all when it comes to great performance.

This is known as the “alignment” problem. It’s discussed at length in other excellent books, like Superintelligence, Stuart Russell’s Human Compatible, and Brian Christian’s The Alignment Problem, so I won’t go into it in depth here.

“Catanzaro was convinced the solution was to redesign the microchip anew. He cofounded the UC-Berkeley Parallel Computing Lab in the mid-2000s, along with several colleagues. There, Catanzaro made a list of existing parallel applications. The business problem, he could see, was that even for the supposedly hungriest customers, the demand for computing power was capped: once you sold an oil prospector a supercomputer, you saturated demand for years. What you needed, Catanzaro figured, was an application that was so hungry for computation that it could never be satisfied. You needed another application like 3D graphics that demanded more computer power once its initial needs were fulfilled. Eventually, Catanzaro deduced what had to be parallel computing’s killer app. “The answer to that was AI,” Catanzaro said. “I came to AI from the bottom up. I came from a circuits perspective. I felt it was just inevitable that AI was the most important computational workload.

TWENTY-TWO: The Fear

“When I shared Jensen’s objections with Bengio, he grew agitated. “Of course, there is no data!” he said. “Humanity hasn’t disappeared yet! Are we going to wait until we have repeated the death of humanity multiple times to decide that, oh, now we have data?!” He made a good point. All the data in the world would not have predicted the breakthrough of AlexNet or the success of the transformer architecture. Twice in ten years, AI had experienced unforeseeable and permanent upgrades to its capabilities. Bengio did not think that the current models posed an immediate threat to human life—but what about the next breakthrough? No one could say what it might bring or when it might happen.

Newton’s Child

“Perhaps we should strive towards mythical resonances in our lives. Among many possible images, a human being can be seen as a tree: we should reach out for more light even as we reach deeper into reality for a more solid hold on the earth. We were not born with one eye, with only one thought in our heads, and with only one direction to travel. When we look out on the world with all its multiplicity of astonishing phenomena, do we see that only one philosophy can contain, explain, and absorb everything? I think not. The universe will always be greater than us. Our mind therefore should be like Keats’ thoroughfare, through which all thoughts can wander. It should also be a great cunning net that can catch the fishes of possibility.