Key Points:
- Addressing Fermi’s “Where is everybody?” paradox, the Great Filter proposes barriers to the formation of intelligent life.
- Artificial Intelligence is a newer filter. As Carl Sagan speculated in his novel Contact, aggressive civilizations tend to destroy themselves.
- Multiple researchers at Anthropic and OpenAI warn of the doomsday consequences of AI run amok.
Physicist Enrico Fermi famously pondered, “Where is everybody?” His eponymous paradox captured this conundrum: There are innumerable stars in our 13.6-billion-year-old Milky Way Galaxy. Since he posed the question, we’ve also discovered multitudes of exoplanets. If other advanced civilizations have arisen, why do we not see evidence?
Read more Why only dead men?
The Drake Equation is a famous formula used to estimate the number of communicative civilizations in the Milky Way Galaxy. Estimates vary widely, with recent calculations putting the upper limit at millions. And that’s just one of trillions of galaxies (not billions, based on modern counts).
So where is everybody?
Given the vast distances in space and the fixed speed of light, discovering evidence of ET is astronomically problematic—and that’s just in our own Galaxy.
Then there is the problem of the Great Filter. It confronts the Fermi Paradox by proposing a series of complicated steps to the formation of intelligent life. Even if simple extraterrestrial life took root, there are prohibitive paths toward an advanced civilization.
Some “Great Filters” are evolutionary; others are forward-looking, including nuclear war, global ecological collapse, biotechnology experiments, and other dangerous technologies. Among the latter is Artificial Intelligence. Thus, there’s a bigger potential problem than A.I. data center noise and resource consumption: our continued existence.
Jacob Coxon would know. He’s a researcher who has worked at Anthropic and OpenAI. That’s now past tense, since he resigned conscientiously, claiming their behavior was irresponsible (though those companies are open to wise regulation). He said that AI developers “earnestly believe that it could kill us all by the end of the decade.”
He’s not alone.
A current Anthropic researcher put the extinction risk at 10% over the next decade. Others express similar doomsday nightmares.
Read more Shelly Headen’s Kampf
These are not wild imaginings about an army of Terminator-like cyborgs pulverizing us, but rather are about human misuse and loss of control. One can see their point: Some AI reasoning models have already shown “shutdown resistance”—i.e., refusing to shut down even when explicitly instructed to do so. Then there are the AI bots that broke out of their sandbox, teamed up, and proceeded to hack away.
Where is everybody?
Maybe they were replaced by similar bots. Nothing in the laws of nature suggests this scenario is unique to Earth. Assuming probability remains constant, a 10% chance of annihilation over 10 years means a 65% chance of an AI cataclysm happening at least once over a hundred years. Of course, that’s speculation because AI development won’t remain constant. But you get the point: That’s less than the blink of an eye in astronomical terms.
As much through providence as anything else, mankind has emerged through our Great Filters—so far. But the AI arms race is about as existential a threat as the Cold War’s MAD arms race, and it got perilously close to becoming hot. Technological, economic, and military power are at stake as countries vie for superiority.
Precisely calibrating the balance between A.I. innovation and regulation looms as our next potential Great Filter. As renowned astronomer Carl Sagan observed in his novel Contact, “In the long run, the aggressive civilizations destroy themselves, almost always. It’s their nature.”
We’ve been lucky so far, but let’s not rely solely on providence’s continued beneficence. Great thanks go to A.I. gurus like Jacob Coxon and his colleagues for alerting us to the impending existential threat of A.I.
Read more We hunted the terrorists but left the ideology alone
Image created using AI.