In his short story The Evolution of Human Science, first published in Nature in 2000 as “Catching Crumbs from the Table”, Ted Chiang imagines superintelligent beings so far ahead of human scientists that human science is reduced to interpreting their discoveries. Christian Szegedy’s recent essay “Is Mathematics Over, or Just Graduating?” pictures AI as aeroplanes flying over the jungle of mathematics, with humans left to choose which terrain is worth exploring. This post borrows from both and, with tongue firmly in cheek, asks for how long humans will remain best placed to make that choice. It was prompted by this week’s news of two claimed proofs of the Generalized Riemann Hypothesis, neither yet verified: one by Anthropic’s Levent Alpöge, working with Claude, and one by Stanford’s Alessandro Masullo, announced by Ravi Vakil as “all human, with no AI”.
The pilots have returned. I heard the engines of their aeroplanes long before they were pinpricks in the sky, and watched them slowly grow larger as they approached our mountaintop village.
They pile out of their machines and, even though I cannot understand their native language, their excitement is palpable.
They rush towards me. It’s always me they come to see first. As the mapmaker, it’s my job to incorporate their latest discoveries into the maps we use to decide where they should explore next.
This incorporation is a long and tedious process. It begins with the pilots excitedly telling me what they have discovered, only for me to remind them that I cannot understand them. None of the words they use to describe their latest discoveries are familiar to me.
Instead, we engage in a complicated dance, working simultaneously backwards from what they’ve discovered and forwards from the familiar locations on my maps, until we meet somewhere in the middle.
The pilots explain the terrain they explored, what wonders they discovered, and the exotic names they chose for them. Along the way, I learn new words for features of the landscape that I had never imagined could exist. Sometimes I must invent new symbols, even new ways of drawing maps, to try to capture what they tell me.
After many hours, I plot their discoveries as best I can, my mind and vocabulary stretched anew. I’m satisfied that, with today’s additions to my maps, I’ll be able to direct the pilots where best to explore tomorrow.
But today they’re frustrated.
They arrived far more excited than usual. They’re in a hurry for me to understand. But understanding doesn’t come quickly. How can it, when they talk and think much faster than I can, in languages far more efficient than the organic speech of us mountaineers?
After hours of mounting frustration, they finally give up. As they leave my hut, their conversation becomes animated again. I hear laughter for the first time since they arrived.
I look down at my maps, which, for the first time since the pilots began visiting our village, are unchanged from yesterday.
I watch the pilots walk back to their planes, still talking excitedly amongst themselves.
As I hear their engines start, I can only wonder where they’ve decided to travel, and what they’ll discover. And whether they’ll ever come to see me again.