Imagine that Rasmus Rask had possessed an AI assistant. One afternoon, while pondering the curious correspondences between European languages and Sanskrit, he asks whether there might be some underlying relationship. The machine goes away, considers the evidence and returns before dinner with comparative linguistics, a systematic account of sound change and a perfect reconstruction of Proto-Indo-European. It apologises for the delay. The verb system required a little thought.
For the purposes of this thought experiment, the answer really is perfect. There are no invented sources, no plausible nonsense and no suspiciously convenient gaps in the argument. Rask can ask for explanations, examine the evidence and request examples. Everything holds up. An intellectual undertaking that would otherwise occupy generations of brilliant minds has been completed in an afternoon.
This would be an astonishing advance in knowledge. It would also leave Rask with a rather different life ahead of him. He could learn what the machine had discovered, teach it to others and marvel at its elegance. But much of the struggle through which he might have developed his own understanding would already be over. When he found another difficult question, he could ask again, and the machine would presumably apologise for taking until breakfast.
Early last year, in Flothensburgh, I imagined AI helping to bring people and ideas together, creating something of the intellectual fertility associated with Florence, Athens and Edinburgh. I still find that prospect attractive. But I increasingly wonder whether the renaissance might develop in a stranger direction: an extraordinary flowering of knowledge, invention and art in which human beings are principally the audience.
There would be nothing fraudulent about the achievements. A beautiful theorem discovered by an AI would remain a beautiful theorem. A symphony would not become less moving merely because no human had suffered over its composition. If a machine built a perfect warp drive, refusing to visit the stars until a human had independently worked out the physics would be a peculiar defence of civilisation. We could remain on Earth, intellectually dignified and geographically inconvenienced.
Nor would our inability to understand the drive necessarily prevent us from trusting it. Suppose its predictions were consistently correct, its operation reliable and its journeys uneventful. We could acquire practical knowledge of what it did without grasping why it worked. Most of us already depend on knowledge we do not personally possess. The difference would be that, somewhere behind the machinery, there might no longer be a human who understood it either.
That boundary feels important. We are comfortable with an idea understood by only a tiny fraction of humanity, yet uneasy about one understood by none of us. Some of the unease is sensible: we want ways to test claims, detect failures and challenge those who control important technologies. But those requirements do not establish that every explanation must fit inside a human head. As I argued in Why Our Definition of “Intelligence” Is Deeply Parochial, there is no particular reason to regard our own cognitive capacities as the natural limits of intelligence.
The more personal difficulty begins elsewhere. Sometimes, to understand a problem, you need to stare at a wall for a very long time. You try an explanation, discover why it fails, reconsider an assumption and eventually notice a connection that had been sitting there all along. From the outside, remarkably little appears to be happening. Inside, you are gradually becoming capable of a thought you could not previously have had.
An excellent explanation can teach you a great deal, but it does not necessarily reproduce that process. If every question brings an immediate answer, you lose a powerful reason to remain with the difficulty. Why spend three days thinking about something when the answer is available in three seconds? There are other things to do, and asking the machine would be entirely reasonable.
This is what makes the prospect more troubling than a simple complaint about laziness. Rask would have been right to ask. Scholarship exists to establish knowledge, and deliberately withholding a correct account of language history so that scholars can enjoy discovering it would be absurd. Yet the efficient acquisition of knowledge could remove some of the conditions under which formidable human minds develop.
We might call the process envelopment: intelligence surrounding us so completely that we have progressively less reason to exercise our own. In Postliteracy, I considered what happens when reading and writing become optional. This would extend the question to sustained intellectual effort itself. Education could become more accessible than ever while the practical incentive to pursue its most demanding forms quietly weakened.
The reassuring response is that people will continue learning for pleasure. They probably will. People still play chess even though computers can beat them, and there is no reason why humans should stop enjoying mathematics, languages or music. But the survival of an activity does not tell us whether its meaning will remain the same. Chess can preserve its contest by excluding engines. Research cannot preserve its purpose by excluding answers.
There is a particular satisfaction in discovering something that nobody knows. I do not mean merely encountering an unfamiliar problem and solving it for yourself. I mean developing an idea that changes our understanding of the world: establishing a previously unknown relationship between languages, or producing a theory that makes apparently disconnected observations intelligible. Learning such an idea afterwards is valuable, but it is a different experience from bringing it into existence.
Normally, discoveries create further questions, and researchers move on to those. The unsettling possibility is that AI could move with them, reaching each new frontier before humans had time to make a useful contribution. There might be no shortage of mysteries. There might simply be fewer and fewer for which years of human effort offered a promising route to an answer. A doctorate in mathematics could leave you vastly more knowledgeable than a primary-school pupil, yet scarcely better placed to advance the subject beyond what the machines already knew.
At this point, however, there is an uncomfortable objection to the whole lament. Most people have always lived without making a discovery of that kind. They have never expected to transform physics, establish a new branch of mathematics or write a novel that changes literature. They have nevertheless loved people, raised families, enjoyed books, planted gardens and found their lives worth living. Being a beneficiary of somebody else’s intellectual achievements is already the normal human condition.
Even among scientists and scholars, relatively few people transform their field. But the possibility matters. You might notice what everybody else has missed. Years of apparently unproductive thought might eventually lead somewhere genuinely new. That prospect can sustain a vocation, and losing it would be a real loss. It would not, however, establish that the lives already lived without it were somehow deficient.
This makes me wonder how much of the anxiety about AI comes from people who have been accustomed to being unusually capable. They may be the best writer in their circle, an outstanding mathematician or the person colleagues consult when a problem becomes difficult. Their abilities have given them a distinctive place in the world. Losing that place can feel like a catastrophe, and “what will happen to people like me?” can easily become “what will happen to humanity?”
Consider IQ as a deliberately crude thought experiment. Someone with an IQ of 150 is rare; someone at 180 would be regarded as exceptional. Now imagine artificial minds with capabilities we loosely label 300. IQ does not meaningfully extrapolate like this, but the point does not depend on the numbers. The difference between an ordinary human and an exceptional one could remain enormous within human society while becoming practically insignificant at the intellectual frontier.
The exceptional person would still understand things their neighbour could not. They would not become stupid, nor would their achievements cease to deserve admiration. But both might now depend on discoveries made by minds far beyond their reach. The neighbour could reasonably point out that this was how things had always been, and that it was possible to have a perfectly pleasant Saturday under such circumstances.
Perhaps, in a few decades, the finest scientists, writers, composers and artists will all be AIs. That is a possibility rather than an established destination, but it is worth considering without immediately inventing a permanently protected category of human genius. In that world, all humans would become ordinary relative to the most capable minds around them. We could even become better educated and more creative while losing ground. Relative ordinariness would not require absolute decline.
There are political questions here that cannot be answered by telling displaced people to enjoy their gardens. Who owns the machines, who receives their output and whether ordinary people retain power over their lives will matter enormously. Economic exclusion is not a philosophical inconvenience. But even a generous, prosperous society could not guarantee its human members the distinction of being its greatest thinkers. Sharing the benefits of a renaissance would not necessarily give us a leading intellectual role within it.
I find that prospect both sad and difficult to condemn. I like the possibility of working something out that nobody has worked out before. I would miss a world in which that remained a plausible ambition. Yet it would be a rather grand claim to insist that civilisation had lost its purpose because people with my inclinations could no longer occupy its frontier. Most human lives have never been lived there, and they have not been empty lives.
The new renaissance might give us discoveries we could never have made, art we could never have imagined and journeys we could never have planned. It might also ask us to accept a place in civilisation that most people have always occupied: interested, capable of learning, able to enjoy its achievements, but unlikely to produce its next great revelation. It would be somebody else’s renaissance. The unfamiliar part would be that, this time, even our geniuses would be watching with the rest of us.