When historians in future centuries compile the complete annals of humankind, their output will be divided into two tomes. The first will cover the hundreds of thousands of years during which humans have been earth’s highest form of intelligence. It will recount how souped-up apes came up with stone tools, writing, sliced bread, nuclear weapons, space travel and the internet—and the various ways they found to misuse them. The second tome will describe how humans coped with a form of intelligence higher than their own. How did our sort fare once we were outsmarted? Rather thrillingly, the opening pages of that second volume may be about to be written. Depending on whom you ask, artificial general intelligence—systems capable of matching humans, and then leaving them in the cognitive dust—are either months, years or a decade or two away. Predictions of how this might pan out range from everyone enjoying a life of leisure to the extinction of the human race at the hands of paperclip-twisting robots.
ChatGPT is but a couple of years old; AI bigwigs disagree whether its intelligence can yet be compared to that of a cat. If that sounds reassuring, consider the newt. Not just any salamander, but rather the fictional heroes of “War with the Newts” by Karel Capek, published in Czechoslovakia in 1936—an excellent beach read and a chillingly prophetic allegory of developments in AI. In the satirical novel, the captain of a Dutch ship stumbles upon a breed of sea creatures in Indonesia. The crew is bemused when the child-sized beasts playfully throw stones back at them, and seem to respond to human prompts. How clever! Just as we were all amused in 2022 by AI’s ability to generate an image of dogs playing poker in the style of Caravaggio, our fictional forebears marvelled at how these fast-learning new forms of intelligence could extract valuable pearls. Then as now, an opportunity to profit is spotted. Soon the newts demand knives so as to produce more pearls. Sure thing, for what could possibly go wrong?
The comedy of unintended consequences that ensues will sound familiar to AI worrywarts. The knives help the salamanders defeat predatory sharks that had been holding their numbers in check, resulting in a demographic boom worthy of Moore’s Law. As the newts multiply, it is clear they are dimmer than humans but evolving at a startling clip. Before long they are conversing as fluently as a chatbot. A nebulous “Salamander Syndicate”, a sort of Big Tech selling newts by the millions, finds new ways to train the plentiful creatures and put them to work in farms and factories. The Panama Canal is expanded at little cost to humans. Fears that this is all happening a little too fast are dismissed as reheated Luddism. The bonanza of cheap worker-newts is decried as “anti-social” by the International Labour Organisation; today, it is the IMF warning that 40% of jobs may be affected by AI. Australia imposes a levy on newts, akin to the tax on robots that Bill Gates, a prolific enabler of machines, once suggested.
Not everyone is happy about this “Salamander era”, a forerunner of today’s “age of AI”. The creatures’ consumption of the global food supply worries critics, much as machine-learning models are now criticised for their electricity usage. Questions are asked about whether humans who hire newts are liable for their actions, just as a tribunal recently forced Air Canada to honour a discount made by an AI-enabled chatbot that had hallucinated the policy. Some people get attached to newts as they learn to dance and sing. The Daily Star asks “Have Newts got a soul?”; in 2023 America, a Catholic magazine, went for “Does ChatGPT have a soul?” Many today feel software is metaphorically eating the world. Capek’s newts instead guzzle land, creating new coastal habitats to host their ever-increasing numbers.
The reassuring newt salesmen argue that their products merely mimic humans and cannot surpass them (this also sounds familiar today). Inevitably, the once-dim creatures eventually do just that. By the time people realise they have no idea how newts actually work—what happens under water is as mysterious to Capek’s humans as what happens in the cloud is to today’s AI users—it is too late. People are too wrapped up in their own obsessions to notice what is going on—the Germans are busy arguing that their newts are ethnically superior to other ones. The newts can always find some greedy human to sell them explosives or submarines.
Newtvidia
Capek was active in Czech political circles and studied philosophy in Paris and Berlin. There is something archetypally European about his fear of the New Thing. There are good reasons for Europe to fret about rapid changes brought forth by fast-evolving newts or computers. Particularly in southern Europe, social systems depend on things evolving at a pace with which the often overbearing state can keep up. In the book the French are the first to try to restrain the newt takeover. Today, even as America does much of the investing in AI, it is the European Union that has spearheaded its regulation. The bloc’s AI Act, thus far the world’s most stringent straitjacket on the technology, will formally come into force in August. Machines that recognise faces in public and Chinese-style “social scoring” systems will be banned.
Capek also imagined a dystopian future at the hands of robots, a term he coined in an earlier work (it is Czech for “serf”). But not even a science-fiction writer could fathom a human-made technology evolving at the pace of today’s AI—hence the newts. In the book, the weak-willed humans hand over China to their new overlords in the hope of buying peace. In 1938, just weeks before Capek died, his homeland was similarly sacrificed to appease Nazis. “War with the Newts” has been praised for its depiction of 20th-century humans’ hubris and greed. It may someday end up being remembered for its understanding of 21st-century machines.