Since the release of ChatGPT last November, apocalyptic warnings that AGI, or artificial general intelligence, could destroy humanity have been all over the news. "AI poses 'risk of extinction,'" says "leaders from OpenAI, Google DeepMind, Anthropic, and other AI labs," the New York Times reported last May. The previous month, TIME magazine published an article by the leading "AI doomer," Eliezer Yudkowsky, who declared that "the most likely result of building a superhumanly smart AI, under anything remotely like the current circumstances, is that literally everyone on Earth will die." Similarly, an AI researcher named Connor Leahy told Christiane Amanpour in an interview for CNN that the prospect of AGI killing off the entire human population was "quite likely."
At the very same time, the prospect of "God-like AI" has also inspired a flurry of utopian proclamations. Tech billionaire Marc Andreessen claims that advanced AI will radically accelerate economic growth and job creation, leading to "heightened material prosperity across the planet." It will also enable us to "profoundly augment human intelligence," cure all diseases and build an interstellar civilization. The CEO of OpenAI, Sam Altman, echoes these promises, arguing that AGI will make space colonization possible, create "unlimited intelligence and energy" and ultimately produce "a world in which humanity flourishes to a degree that is probably impossible for any of us to fully visualize yet."
All of this might seem unprecedented. There are so many dire warnings of imminent extinction in the news right now, sometimes paired with equally wild predictions that a new era of radical abundance lies just around the corner. Surely something big is happening. Yet this isn't the first time that notable scientists and self-described "experts" have announced to the public that we're on the cusp of creating a magical new technology that will either annihilate humanity or usher in a utopian world of unfathomable wonders. We've been here before, and what happened? In every case, the outcome was much less sensational than people were led to believe. Often, the hype turned out to be a giant nothingburger.
To put the frenzied hype around AGI into historical perspective, let's revisit one such episode from the early 20th century. Understanding that history will demonstrate that what we're seeing now is nothing new.
It began with the discovery of radioactivity in 1896 by the French physicist Henri Becquerel. What is radioactivity? Let's start by imagining that you place a chunk of iron in direct sunlight for a few hours and then move it to a dark room. If you touch the iron right after moving it inside, it will feel pretty hot, right? But with each passing minute its temperature will drop, until it returns to room temperature.
This is simple enough: The iron rod absorbed energy from the sun and then re-radiated it in the form of thermal energy, which we experience as heat. Without the sunlight — an external source of energy — the temperature of the rod will equilibrate to the temperature of its environment.
Now let's imagine a different chunk of metal. We place it in a dark, cool room for several days, only to discover that it's actually radiating energy on its own. That's what Becquerel found: The metal called uranium will give off a slight glow even if it's kept in a dark room with no external source of energy. This glow can't be seen with the naked eye, but if you place the uranium next to a photographic plate, an image of it will appear even if the uranium has been stored in a pitch-black room for weeks at a time. How can it radiate energy without an external source?
Becquerel's observation didn't get much attention at first. That all changed after Marie Curie discovered that radium, another type of metal, also produced energy on its own — but in much greater quantities. In fact, you can literally see radium glowing with the naked eye in a dark room. Curie coined the word "radioactivity" to denote this phenomenon, though she had no idea how or why it was happening. A metal that could produce its own internal energy at first seemed like a violation of the laws of physics.
An explanation finally came in 1901 from a pair of physicists, Frederick Soddy and Ernest Rutherford. Their discovery was mind-blowing: Some atoms in the radioactive metal spontaneously turned into atoms of a completely different kind of metal, and each time that happened, a small amount of energy was released. That's how these metals produce energy without an external source: Uranium atoms, one at a time, morph into atoms of a different metal, thorium, through a process called radioactive decay. Atoms of thorium, which is also radioactive, then decay into other types of atoms, including radium, until the entire clump becomes a "stable" — that is, non-radioactive — form of lead, the heavy metal formerly used in paint and gasoline. That ends the process of radioactive decay, which has produced energy from beginning to end.
.svg){kind=link}
What physicists Soddy and Rutherford realized was that nature itself is an alchemist, "transmuting" materials into other types of materials through the spontaneous process of radioactive decay.
In previous centuries, alchemists had tried to convert one type of metal into another, usually lead into gold, with a notable lack of success. What Soddy and Rutherford realized was that nature itself is an alchemist, "transmuting" materials into other types of materials through the spontaneous process of radioactive decay. Indeed, when Soddy realized what was going on, he shouted to his colleague: "Rutherford, this is transmutation!" Rutherford then shot back: "For Mike's sake, Soddy, don't call it transmutation. They'll have our heads off as alchemists." Alchemy had long since lost any respectability among professional scientists, and Rutherford didn't want to jeopardize their careers.
An even more significant discovery happened a year later, in 1902, when Soddy and Rutherford found that the amount of energy produced by radioactive decay was enormous — not in "absolute" terms but "relative" to the size of the atoms. As historian Spencer Weart writes, the duo's research "showed that radioactivity released vastly more energy, atom for atom, than any other process known."
Exactly how much energy does radioactive decay produce? The answer is given by Albert Einstein's famous equation E=mc2, first published in a 1905 paper that introduced his "theory of special relativity."
That equation says two important things about the peculiar nature of our universe: First, it states that mass and energy are equivalent. They are "different manifestations of the same thing," as Einstein explained in a 1948 interview. No one at the time believed that — mass and energy were clearly different types of phenomena, it was assumed, but Einstein showed that this commonsense intuitive idea was wrong.
Second, the equation states that small amounts of mass are equal to enormous amounts of energy. To calculate the amount of energy contained in some quantity of mass, you first square the "c," which stands for the speed of light (a very large number), and then multiply the resulting number — the c2 — by the amount of mass in question. The result is the amount of energy you get if that mass is converted into energy. In Einstein's words, the E=mc2 equation shows "that very small amounts of mass may be converted into very large amounts of energy."
We need your help to stay independent
This means that atoms contain a colossal storehouse of energy — "atomic energy," as it was called at first, although "nuclear energy" is more common today. This atomic energy is what radioactive materials give off when they spontaneously decay: As the atoms of one type of metal transmute into atoms of another, they lose a little bit of mass, and this lost mass is converted into energy. That's how radioactive metals like uranium and radium produce their own internal energy, without any external source.
The implications of this extraordinary discovery were profound. If there were some way to extract, harness or liberate this great reservoir of atomic energy, then tiny amounts of mass could be used to power entire civilizations. Atomic energy could usher in a new era of endless abundance, a post-scarcity world in which the energy available to us would be virtually "inexhaustible." As Soddy declared in a popular book published in 1908,
A race which could transmute matter would have little need to earn its bread by the sweat of its brow. If we can judge from what our engineers accomplish with their comparatively restricted supplies of energy, such a race could transform a desert continent, thaw the frozen poles, and make the whole world one smiling Garden of Eden. Possibly they could explore the outer realms of space, emigrating to more favourable worlds as the superfluous to-day emigrate to more favourable continents.
Elsewhere he claimed that, by releasing the energy stored in atoms, "the future would bear as little relation to the past as the life of a dragonfly does to that of its aquatic prototype," and that "a pint bottle of uranium contained enough energy to drive an ocean liner from London to Sydney and back."
One prominent scientist prophesied that nuclear energy would "almost instantaneously change the face of the world ... the poor will be equal to the rich and there will be no more social problems."
Journalists ate all this up, raving about the transformative potential of atomic energy on the pages of leading newspapers and magazines. "When Rutherford and Soddy pointed out that radioactive forces might be the long-sought source of the sun's own energy," Weart writes, "the press took up the idea with relish. Instead of sustaining future civilization with solar steam boilers, perhaps scientists would create solar energy itself in a bottle!" One of the most prominent scientific voices of his day, Gustave Le Bon, prophesied that "the scientist who finds the means of economically releasing the forces contained in matter will almost instantaneously change the face of the world," adding that "the poor will be equal to the rich and there will be no more social problems."
By the 1920s, most people — including many schoolchildren — were familiar with the idea that atomic energy would revolutionize society. Some even predicted that controlled transmutation might produce gold as an accidental by-product, which could make people rich while solving all our energy woes. Exemplifying hopes that a Golden Age lay just ahead, Waldemar Kaempffert wrote in a 1934 New York Times article that although we couldn't yet unlock the storehouse of energy in atoms, a method would soon be discovered, and once that happened, "probably one building no larger than a small-town postoffice of our time will contain all the apparatus required to obtain enough atomic energy for the entire United States."
This was the utopian side of the hype around radioactivity. Yet just as sensational were the apocalyptic cries that the very same phenomenon could destroy the world — and perhaps even the entire universe. In 1903, two years after discovering transmutation, Soddy described our planetary home as "a storehouse stuffed with explosives, inconceivably more powerful than any we know of, and possibly only awaiting a suitable detonator to cause the earth to revert to chaos." Le Bon worried about a device that, with the push of a button, could "blow up the whole earth." Similarly, in a 1904 book, scientist and historian William Cecil Dampier wrote that
it is conceivable that some means may one day be found for inducing radio-active change in elements which are not normally subject to it. Professor Rutherford has playfully suggested to [me] the disquieting idea that, could a proper detonator be discovered, an explosive wave of atomic disintegration might be started through all matter which would transmute the whole mass of the globe into helium or similar gases.
This is the idea of a planetary chain reaction: a process of contagious radioactivity, whereby the decay of one type of atom triggers the decay of other atoms in its vicinity, until the entire earth has been reduced to a ghostly puff of gas. Human civilization would be obliterated.
Some even linked this possibility with novae observed in the sky — sudden bursts of light that dazzle the midnight firmament. What if these novae were actually the remnants of technological civilizations like ours, which had in fact discovered the dreaded "detonator" referenced by Rutherford? What if novae were, as one textbook put it, "brought about perhaps by the 'super-wisdom' [i.e., the technological capabilities] of the unlucky inhabitants themselves?"
This was not a fringe idea. Frédéric Joliot-Curie, the son-in-law of Marie Curie, even mentioned it in his Nobel Prize speech, delivered in 1935 after he and his wife, Irène, discovered a way to cause radioactive decay to occur in otherwise non-radioactive materials, a phenomenon known as artificial radioactivity. "If such transmutations do succeed in spreading in matter," Joliot-Curie declared to his Nobel audience,
the enormous liberation of usable energy can be imagined. But, unfortunately, if the contagion spreads to all the elements of our planet, the consequences of unloosing such a cataclysm can only be viewed with apprehension. Astronomers sometimes observe that a star of medium magnitude increases suddenly in size; a star invisible to the naked eye may become very brilliant and visible without any telescope — the appearance of a Nova. This sudden flaring up of the star is perhaps due to transmutations of an explosive character like those which our wandering imagination is perceiving now — a process that the investigators will no doubt attempt to realize while taking, we hope, the necessary precautions.
At the extreme, some even reported to the public that "eminent scientists" thought this chain reaction of radioactive decay might spread throughout the universe as a whole, destroying not just our planet but the entire cosmos. By the 1930s, Weart notes, "even schoolchildren had heard about the risk of a runaway atomic experiment."
Or perhaps radioactivity would bring about a dystopian nightmare: As Rutherford liked to say, "Some fool in a laboratory might blow up the universe unawares" by triggering a planetary chain reaction.
These were the grandiose promises and existential fears associated with radioactivity. They were promulgated by leading scientists, amplified by the media and so widely discussed that even children became familiar with them. What lay ahead, people were told, was a utopian world of limitless energy in which all societal problems will be solved. Or, on the other hand, radioactivity could bring about a dystopian nightmare in which, as Rutherford liked to say, "some fool in a laboratory might blow up the universe unawares" by inadvertently triggering a planetary chain reaction through some artificial radioactivity process.
The parallels with the current hype around AGI are striking. Today, one finds prominent figures like Andreessen and Altman proclaiming that AGI could solve virtually all our problems, ushering in a utopian world of "heightened material prosperity across the planet," "unlimited intelligence and energy" and human flourishing "to a degree that is probably impossible for any of us to fully visualize yet."
At the same time, Altman notes that the worst-case outcome of AGI could be "lights-out for all of us," meaning total human extinction, caused not by a planetary chain reaction but by a different exponential process called "recursive self-improvement," which some believe could trigger an "intelligence explosion." These doomsday prophecies have been further amplified by AI researchers like Geoffrey Hinton and Yoshua Bengio, both of whom won the Turing Award, often called the "Nobel Prize of Computing."
Meanwhile, the media has lapped up all this hype, both utopian and apocalyptic, amplifying these warnings of existential doom while also declaring that AGI could revolutionize our world for the better.
Want a daily wrap-up of all the news and commentary Salon has to offer? Subscribe to our morning newsletter, Crash Course.
Historians of science and technology have seen this all before. The details were different, but the hype wasn't. If the past is any guide to the future, the push to create AGI by building ever-larger "language models" — the systems that power ChatGPT and other chatbots — will end up a giant nothingburger despite the grand proclamations all over the media.
Furthermore, there is another important parallel between radioactivity in the early 20th century and the current race to create AGI. This was pointed out to me by Beth Singler, an anthropologist who studies the links between AI and religion at the University of Zurich. She notes that just as the dangers of the everyday uses of radioactivity were ignored, the harmful everyday uses of AI are being ignored in public discourse in favor of the potential AI apocalypse.
Not long after Marie Curie wowed audiences at a major scientific conference in 1900 with vials of radium "so active that they glowed with a pearly light," a physician who studied radioactivity with Marie Curie, Sabin Arnold von Sochocky, realized that adding radium to paint caused the paint to glow in the dark. He co-founded a company that began to manufacture this paint, which was used to illuminate aircraft instruments, compasses and watches. It proved especially useful during World War I, when soldiers began to fasten their pocket watches to their wrists and needed a way to see the time in the dark trenches to synchronize their movements.
Exposure to the gamma rays emitted by radium poses a radiological hazard, however, which very likely caused Sochocky's own death at age 45. Worse, as Singler points out, throughout the 1910s and 1920s many women who painted these watches in factories owned by Sochocky and others came down with radiation poisoning; some died and others became extremely ill. Some, such as Amelia Maggia, died after suffering a number of horrendous health complications. Several months after Maggia quit her dial-painting job, "her lower jawbone and the surrounding tissue had so deteriorated that her dentist lifted her entire mandible out of her mouth." She passed away shortly after that.
The victims of this new industry, the women poisoned or killed who were known as the "radium girls," were collateral damage of the push to get rich off radioactivity.
The victims of this industry were called the "radium girls," as most factory workers were young women. They were the unwitting collateral damage of a push by Sochocky and others to get rich off the hype surrounding radium. In reality, the radium industry both generated huge profits and caused great harm, leaving many workers with devastating illnesses and killing many others.
Similar points can be made about the race to create AGI. Lost in the cacophony of grand promises and apocalyptic warnings are myriad harms affecting artists, writers, workers in the Global South and marginalized communities.
For example, in building systems like ChatGPT, OpenAI hired a company that paid Kenyan workers as little as $1.32 per hour to sift through some of the darkest corners of the web. This included "examples of violence, hate speech, and sexual abuse," leaving many workers traumatized and without proper mental health care. OpenAI also used, without permission, attribution or compensation, an enormous amount of material generated by human writers and artists, which has resulted in lawsuits for intellectual property theft that are now going to court. Meanwhile, AI systems like ChatGPT are already taking people's jobs, and some worry about widespread unemployment as OpenAI and other companies develop more advanced AI programs.
While some of this has been reported by the media, it hasn't received nearly as much coverage as the dire warnings that AGI is right around the corner, and that once it arrives, it may kill everyone on Earth. Just as the rush to cash in on radium destroyed people's lives, so too is the race to build AGI leaving a trail of damage and destruction.
The lesson here is twofold: First, we should be skeptical of claims that AGI will either bring about a utopian paradise or annihilate humanity, as scientists and crackpots alike have made identical claims in the past. And second, we must not overlook the many profound harms that AGI hype tends to obscure. If I had to guess, I'd say that AGI is the new radium, that the bubble will burst soon enough, and that companies like OpenAI will have achieved little more than hurting innocent people in the process.
Read more
from Émile P. Torres on the AI revolution
Shares