Nobody Knows the Future

A t the turn of this millennium, the titans of the tech industry were breathless with anticipation for a revolutionary new product. They referred to it mysteriously by code names like “Ginger” and “IT.” Based on a leaked book proposal, Time magazine reported that “IT” was expected to change the world by those most firmly in the know: “As big a deal as the PC, said Steve Jobs; maybe bigger than the Internet, said John Doerr, the venture capitalist behind Netscape, Amazon.com and now Ginger.” Jeff Bezos called it “one of the most famous and anticipated product introductions of all time.”

The product, unveiled in December of 2001 on Good Morning America, was the Segway: a motorized two-wheeled scooter. Readers today may vaguely remember it as the preferred conveyance of Paul Blart, Mall Cop. The Segway was discontinued in 2020, and in the meantime, it did not redesign cities or render cars obsolete, as was foretold.

On the other hand, the Segway wasn’t totally without merit: The gyroscope technology inside it enabled other novel solutions to many personal transportation problems, especially for the disabled. Electric scooters did arguably end up changing urban landscapes, to a more modest extent, with the advent of services like Lyft and Bird.

The point here is not that inventors are harebrained buffoons or that their aspirations are silly. Instead, the Segway reminds us of something far simpler: Nobody knows the future. This is age-old but also radically challenging wisdom: “Who can tell someone else what is to come?” asks the book of Ecclesiastes, rhetorically. The classicist Armand D’Angour, in his reading the archaic poet Hesiod, writes, “The lexical evidence suggests that the Greeks of Hesiod’s time thought of the past as in front of our eyes and the future as behind our backs.” In other words, what is to come is by definition what we can’t yet see.

In our own present, however, science has given us such rigorous predictive power when it comes to the physical world that it’s tempting to believe we can game out every variable at will. This is probably what accounts for the confident pronouncements that business magnates and secular prophets still make about what “science” should lead us to expect.

For instance: It was March of 2025 when Dario Amodei, CEO of the AI company Anthropic, said that “I think we will be there in three to six months, where AI is writing 90 percent of the code. And then, in 12 months, we may be in a world where AI is writing essentially all of the code.” In May, per Axios, he was warning that “AI could wipe out half of all entry-level white-collar jobs — and spike unemployment to 10–20 percent in the next one to five years.”

That was more than a year ago, and AI is now writing a fair amount of code. But by no conceivable estimate does its output account for 90 percent of the total, whatever that would mean. Attaching a number to these sorts of pronouncements has a way of making them seem intellectually precise, but it’s hard to say how anyone could tabulate either the numerator or the denominator of that percentage.

How much code is written in the world, exactly, and how much is written by AI? In the world of mathematical abstraction, it might be possible to imagine that an answer exists. In the real world, where programmers typically prompt Large Language Models (LLMs) to produce code but then refine their output and work in tandem with them on revision, how could anyone tease out the machine’s contributions from the human’s well enough to attach a reliable figure to each? Even in relatively computational problems like this one, the numerically minded have a habit of speaking as if they are much more certain than anyone can be.

What’s more, outside the domain of mathematics, some things might be fundamentally irreducible to the satisfying absolutes of numbers. This may now count as heresy in certain quarters, but it’s what the ancients knew: “Our discussion will be adequate if it has as much clarity as the subject matter can sustain,” wrote Aristotle in his Nicomachean Ethics. Politics, morality, and human action “admit of much dispute and variability.” They are fundamentally unpredictable: It is as silly to apply mathematical proof to an ethical deliberation as it is to vote in Congress on the surface area of a cube.

Perhaps this is why the forecasts of our most supposedly hyper-rationalist intellectuals keep falling so embarrassingly flat, and why they persist in making new ones just the same. In 1992, Al Gore wrote that “up to 60 percent of the present population of Florida may have to be relocated” — according to some predictions — “in the next few decades.” Florida has remained firmly in place, but that didn’t stop Al Gore from making equally self-assured — and wrong — declarations in his 2006 documentary, An Inconvenient Truth.

AI is not a Segway, and Dario Amodei is by no means as far off the mark as Al Gore. Deep learning technology has already revolutionized many industries, and its effects on us — for good and ill — will likely continue to be profound. Now that AI leaders across the industry are raising concerns about an imminent spike in the dangers and capacities of the technology, it makes sense to take those concerns seriously.

But not to take them as gospel. It would lower the temperature of the current AI conversation substantially if we could recall that neither Amodei, nor Sam Altman, nor Elon Musk, nor you, nor I — no, not even the angels, but only the Father in Heaven — know the future with anything like mathematical certainty.

This is America, after all, and we don’t govern ourselves by the certainties of very intelligent people, even if those certainties come with equations attached. Choosing a reasonable and wise path forward is not merely a matter of crunching numbers but of public debate, moral reasoning, and level-headed deliberation based on our best current knowledge and inherited wisdom. Anything more or less than that is beneath us.