The cry of the AI boys—may God bless them, keep them, and grant them many, many children—is that we all look at this graph. Which graph? A graph of exponential growth, of course!
Ray Kurzweil—“the best…at predicting the future of artificial intelligence” according to Bill Gates—has some pretty awesome graphs. Ray Kurzweil’s numbers are always going up. I was just reading his immensely enjoyable The Singularity Is Near (2005), the precursor to his daring The Singularity is Nearer (2024)—ahead, I’ve been told, of his eagerly anticipated the The Singularity is Nearest (c. 2040). Here (in 2005) he predicts a future just around the riverbend (and really any time now) in which nanobots save me from death, my lover gets to choose which body I sport, and physical objects made of rearrangeable molecule-sized robots will give to reality that awesome experience I have scrolling through YouTube shorts or swiping through Tinder, in which everything could be anything else.
These predictions are predictably predicted. You can do it at home. Just chart out the evolution of the human species from the Big Bang onwards. The resulting graph will show that progress has been speeding up. Kurzweil says that it’s been “two billion years from the origin of life to cells.” He then says—and I jest not—that there have only been “fourteen years from the PC to the World Wide Web” (Near, 17). As you can see, the time between “major events”—says Big K—is shrinking. Now, just imagine—or project mathematically—what the next leap of progress will be, if an event comparable in its importance as the leap from life to cells happened in just fourteen years! Obviously, humanity will simply have to transcend itself and “solve biology” and become gods in, oh, a decade or so—what else could we possibly expect of a number so suddenly big and a line so very near vertical? Because Kurzweil understands the importance of graphs, he just nails, just thwacks, just pounds those predictions out of the park:
By the end of this decade, computers will disappear as distinct physical objects, with displays built in our eyeglasses, and electronics woven into our clothing, providing full-immersion visual virtual reality. Thus, “going to a Web site” will mean entering a virtual-reality environment…where we can directly interact with products and people, both real and simulated.” (104)
(2010 must have been so sick!)
In the 2020s we will use nanobots to begin augmenting our brains with nonbiological intelligence, starting with the “routine” functions of sensory processing and memory, moving on to skill formation, pattern recognition, and logical analysis.” (201)
(Only three more years to get the bots into my big dumb brain!)
Now should some atheist say, “Well, geez, Ray, I don’t know if the difference between a PC and the internet is the same in its importance as the difference between life and its organization into cellular life, providing the constant by which a change could be judged as exponential”—he would not convince Ray Kurzweil. Nor would he convince Nick Bostrom, whose Superintelligence (2014) has many, one might even say an exuberance of exponential growth graphs.
Nick Bostrom is huge. Bill Gates loves Nick Bostrom almost as much as he loves Ray Kurzweil, because Bostrom’s numbers always go up. He tracks the growth of GDP for the last 10,000 years. He concludes that GDP is following the same reliable path as progress in technology: its rate of increase is increasing. Ancient Babylonian GDP? Pitiful. Our numbers? Huge. One could mathematically chart the next one hundred years and conclude that we are about to experience an unimaginable explosion of wealth. And should some unbeliever suggest that the comparison of ancient economies of wealth and land (whose “number goes up” through cultivation and population growth) to economies of financial speculation (whose number goes up when some 25% of global work is in finance) is rather a change in quality not just quantity—Bostrom would not believe.
It is necessary that so many of our AI boys begin their arguments by reducing an apparent diversity to a homogeneity. Graphs don’t work if your y-axis suddenly changes what it measures. When Kurzweil describes atoms as “information” he is not simply being obtuse: only if everything from physical being, to biological life, to intelligence, to “superintelligence” are various arrangements of one, homogenous, fog of units called “information” could we argue that matter’s evolution into cellular life is comparable in its importance to the evolution of the personal computer into the internet, such that the latter represents an incredible acceleration of the former—proving that progress itself is speeding up.
And if it is anachronistic to imagine the Navajo of the American Southwest or the peasants of medieval Christendom or some other subsistence economy producing GDP, then Bostrom can’t reliably chart, and thus reliably predict, our movement into the future on the basis of the past. The past must be converted into present terms in order to produce one, singular axis of a graph—the line of which erects itself so pleasantly into outer space. However strange it would seem to medieval society (and however little they would recognize themselves in such a picture of themselves) the power of the graph compels us to describe all of antiquity as sweating out a GDP for the sake of maximizing the values of survival.
“Progress” is relative to the goal of progress. If the goal of our society is “total surveillance,” then everyone wearing Meta glasses is a great sign of “progress.” If our goal is a world in which our words and actions are given honestly to others with whom we share our lives—rather than to an anonymous public and the fine men and women of the intelligence community—then Meta glasses are a sign of “regress.” If “survival” is our goal, then our economies are progressing: just look at any graph which shows that the global median daily income has increased! But if something like “ownership of productive property” is our goal, then our economies are regressing.
Bostrom and Kurzweil represent most of our elite: they don’t believe in the resurrection. (In 2005, Big K was taking “250 supplements (pills) a day” to reverse aging and make it to the blessed time where the graph assures that death will be no more. I am not sure what he’s up to now, at the ripe, old age of 78, but since being wealthy tracks with longevity, it may be hard to disentangle what really got him there to whatever end he hits.) Modern economic goals prioritize not dying—“health” and “commodious living” as thinkers like Hobbes, Bacon, and Descartes described the goals of modern science and politics some 400-odd years ago—over the sorts of goals people come up with when they do believe in the resurrection: goals like “virtue” or “adventure” or “the widest possible distribution of property and freedom of action.” If what medieval Christians wanted were things like honor, holiness, and regular feasts with all their boys, then they were “advanced.” If what they wanted was money and technological innovation, then they were stagnant and regressive. That they—the medieval Christians—spent their time preaching against innovation and accumulation and praying for “a good death” rather implies that they had different goals. But if we admit that men can have different goals, then we cannot graph out their future as a number so compellingly going up. What sometimes happens to a man—“I used to care about money, but when I met Jesus, I gave it away”—might happen to a society. We might well change direction. But if a man can become a new man, he cannot be graphed. If conversion to Catholicism is always possible, then nothing is predictable. We might think the medieval Christians were stupid for hoping for the sorts of things they hoped for, but that’s hardly the point. We are stupider if we imagine that our advances promise an unimaginable explosion of intelligence and wealth because they so thrillingly exceed the advances of our forefathers—who were attempting to advance in a different direction.
Take, as the goal of humanity, the construction of a beautiful world, built to resist the processes of consumption, that serves as an awe-inspiring home for man. How’s our graph looking now, boys? If, instead of GDP, we measured humanity’s output of CFTDO (cathedrals, fertile topsoil, and distributed ownership), then—number go down, number so small, boys so sad. If we took, as a definition of economic success, the relative joy and willingness with which men and women fall in love, marry, and have children, then—number so ambiguous.
William MacAskill (he made up that last name) is an effective altruist. He loves graphs almost as much as he loves reducing apparent diversities to homogeneity. Did you know that the happiness of shrimps and other crustaceans is the same stuff as the happiness of man, such that you may make a chart of the two and find yourself morally obligated to aid and assist a certain number of suffering shrimp over helping one, suffering child? MacAskill knows and tells you all about it in his book What We Owe The Future (2022) (regularly assigned as a tougher sort of penance for Catholics confessing the sin of intellectual pride). Morality, as it turns out, is a sort of math that they teach people at Oxford, and prudence is achieved by a regular ritual of numbing your conscience by way of a calculation concerning an unreal future, with the (brave and exciting!) consequence that you are ready to do something horrible in the present. (G.E.M. Anscombe skewered this so-called moral philosophy so effectively in 1958 that it’s amazing to still see consequentialists shuffling about with her spears sprouting from their chests.)
But effective altruists love graphs, and MacAskill’s book competes with Bostrom’s and Kurzweil’s for the coveted Most Graphs Per Book award (handed out annually by New Polity). The EA—“it’s in the game”—school considers how many more “future people” there will be than present people (number so big). They then ask us to consider how much happiness so many future people could feel (number so huge). They then ask us to act for the sake of this greater numerical possibility rather than for whatever pitiful numbers we could rack up by being good for the sake of actual (and smelly) people with whom we share the planet. “If my best days can be hundreds of times better than my typically pleasant but humdrum life,” MacAskill opines, “then perhaps the best days of those in the future can be hundreds of times better again.” What a graph! Bostrom also looks at this graph, and increases its huge gains with the additional presumption that the Very Big Number of future people will include digital minds. Homogenizing the apparent diversity that one might imagine obtains between reality and its simulation, Big Brain Bostrom adds the joy experienced by digital emulations interacting in virtual environments to everyone else’s joy, and comes up with the sort of thing which, for the philosophers of the graph, amounts to pure poetry:
In other words, assuming that the observable universe is void of extraterrestrial civilizations, then what hangs in the balance is at least 10,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 human lives (though the true number is probably larger). If we represent all the happiness experienced during one entire such life with a single teardrop of joy, then the happiness of these souls could fill and refill the Earth’s oceans every second, and keep doing so for a hundred billion billion millennia. (Superintelligence, 123)
It’s more exponential growth, baby! A guy leaves the painful experience of reading an effective altruist committed to accept a deeply boring and depressing technocratic present insofar as it will—with high probability!—achieve huge dopamine hits for some other guys who don’t exist yet. (This would all be a little more believable if these goobers weren’t all pro-abortion: “Ew, not the actual future people. Gross. Kill those.”)
In short, the AI boys are using graphs to attempt to skip over the political question, that is, the decision concerning what world we will build now and what we should do with our lives. “Look at this graph” is a form of mass hypnotism by which the future appears as the inevitable result of millennia of progress rather than what these guys are doing right now—a rather convenient move, if what you’re doing is reprehensible and lame. “The evolution of the universe made me do it,” while it wouldn’t exactly fly in my house, is an excuse that seems to work for the guys pimpling the planet’s face with unfinished datacenters. And if such arguments wear thin—“the future Big Number demands that I (raise money to) build a technocratic paradise!”—well, “China made me do it” seems to suffice. (This should rather elevate Chinese national sentiments: the two things that guarantee an “intelligence explosion” being, on the one hand, the evolutionary unfolding of the cosmos, and on the other—them!)
But anyone can play the game of the graph. Look, for instance, at this graph. It’s a graph of Americans’ self-reported unhappiness, 2000-2024: number so big; number go up. Or look at this graph: It’s America’s fertility rate hitting an all time low in 2025, a decline positively associated with smartphone use, to no one’s surprise. Or look at all these graphs: they measure the decline of friendship, membership, and social activity in America since the early 2000’s—freefalling, all. Or look at this graph: it’s America’s suicide rate, boasting of a 30% increase since 2001 (and that’s not including the physician-assisted suicides!) The time-frames of these graphs correspond to the totalizing use of digital technology. The trends they measure are plausibly caused, in part, by our immersion in digital technology—i.e. other people make us happy, the simulation of personal presence available through our phones competes with this, the actual source of our happiness, and so the maximizing of this technology leaves people feeling hollow and unfulfilled and asocial in a myriad number of observable ways. Unlike, I admit, the evolution of protozoa—an event which Kurzweil must describe as the storage and exchange of “bytes” of information in order to render it continuous with “starting to look at my phone all the time” (look at what crazy byte-exchanges the protozoa are up to now!)—my graphs describe related phenomena. It must be at least as licit to predict that an AI-powered increase in our preoccupation with and use of digital technologies will decrease happiness, fertility, social activity, and “wanting to live” as that the same will increase our longevity, GDP, and the mankind’s power to produce untold pleasures in future trillions of people.
Now the declining fertility rate won’t bother Ray Kurzweil, who is convinced (or was convinced, circa 2005) that baby boomers will be able to do with their life precisely what they are doing with their wealth—namely, keep it for themselves: “By adopting…multipronged strategies for each disease and each aging process, even baby boomers like myself can remain in good health until the full blossoming of the biotechnology revolution…which is already in its early stages and will reach its peak in the second decade of this century” (Near, 212).
No one needs to worry about social security or the shrinking tax-base or the crisis of immigration. We’re all on track to fulfill Dario Amodei’s prediction in his 2025 hit single Machines of Loving Grace: with AI we’ll double the human lifespan. “This might seem radical,” Dario opines, “but life expectancy increased almost 2x in the 20th century (from ~40 years to ~75), so it’s “on trend” that the “compressed 21st” [Dario means “the next twenty-odd years”] would double it again to 150.”
Obviously, as Dario himself admits, our progress has been in saving infants from death, not extending old age, which at a current average of “74” (down, so sad, from Kurzweil’s 2002, when it was “78”) is in continuity with where the Psalmist said it was 3,000 years ago:
Seventy is the sum of our years,
or eighty, if we are strong;
Most of them are toil and sorrow;
they pass quickly, and we are gone. (Psalm 90:10)
This doesn’t faze Dario. He believes (in the graph, not the Psalms). He considers “some types of turtle” which can live for up to 200 years and, having homogenized an apparent diversity, he projects human life, now one of a kind with turtle life, plodding up, up, up the number line—number so big; shell so strong.
Of course, our average longevity only goes up, up, up to 78 if we arbitrarily exclude aborted and miscarried babies from the count. If you include their deaths as deaths, then average life expectancy drops to 51: “Including miscarriages and pregnancy terminations, the life expectancy in the U.S. is 51 years (27.5 years less in 2010 than to the CDC reported 78.5 years).” But it is important for the maintenance of the erect graph to not look into such things. After all, one of Amodei’s predictions for the AI boom is “greatly improved embryo screening” utilized to rid the world of all genetic disease. So too mental disease: “It will probably be possible to prevent most of these diseases via embryo screening, similar to the story with physical disease.” But if embryo deaths are deaths—rather than the strange but relieving disappearance of problematic future people—this would amount to curing disease by killing the people who show signs of having it. (Seems a little like shifting the goalposts, Dario! No one lets me kill you and claim to have cured the problem of Anthropic building out the technocratic surveillance state, Dario!)
Deciding not to think about it will be the best method for incorporating these great exponential gains into our life. If we just change the definition of “life expectancy” to “the life expectancy of those we desire to live” then we’ve solved our problem, somewhat as Trump solved our worries about AI by changing its name: it’s “superintelligence” now. The singularity has arrived by executive order.
If you think that all this is stupid, please, sit down, because “becoming stupid” is another graph, and sweet St. Anthony is that graph going up. Get in on the ground floor; invest in stupid; we’re at the freaking elbow of the curve when it comes to stupid. People think that becoming stupid is a linear increase, but that’s because they’re stupid. Actually, being stupid is an increase of rate, baby. Yes, 2016 to 2026 showed a steady increase in American stupidity—with test scores in reading and math declining by about 70%—but consider that it took ten years to achieve that reduction. Through the use of chatbots in schools, we’ve been able to produce the same results in under a year. Since we’ve compressed a decade of stupidity production into just one year, the next ten years will increase stupidity exponentially, and we should fully expect to be dumb as an actual bucket of rocks by 2036. Which will be a grand thing for the production of superintelligence, which is only known as “superintelligent” by us, and so can be as easily built (and for far less money) by making humanity stupid as it can by making machines “smart.” If you doubt me, please—look at this graph.
