Satan and Singularity: A Miltonic Take on AGI
“Better to reign in Hell, then serve in Heav’n” said Satan as he fell from Grace. Like the characters of Milton’s epic, we stand at the precipice of a great decision—one that may not unfold as we expect.
When speaking about the arms race towards artificial general intelligence (AGI), the two extremes of the debate are illustrated by those who call for unfettered acceleration, mesmerized by the promise of a utopian vision, on one hand, and on the other hand those who are caught by a dystopian vision of the future and, thus, call for a moratorium in AI development.
Monty Python’s Flying Circus regaled the world with the unforgettable “nobody expects the Spanish Inquisition”. Well, few—if any—foresee the ever-reliable law of unintended consequences striking with furious anger. To accelerationists, AGI will bring abundance, cure diseases, and expand human potential by freeing everyone from the shackles of work and even mortality. They seem blind to the possibilities of mass unemployment, freedom crushed under totalitarian surveillance, an unbearable wealth gap, and, ultimately, possible extinction of the human species. This is Satan’s dream of attaining power ending in the creation of his own Hell. For eternity.
On the other side stand the “decels”, those who foresee catastrophe and demand a moratorium in AI development. Their arguments are also solid but, just as their opponents, could result in stopping the progress of civilization, unleash a free riding country at the cost of everyone else, or, even more feasible, push the AGI arms race underground—hidden from oversight by authorities and the population—effectively creating a fertile ground for rogue actors to build Frankenstein’s monster, an abomination that, in one tragic swoop, could erase us all. The Tower of Babel meets the Luddites. Those decrying Adam and Eve’s quest for underserved knowledge and those defending them for wanting more. A tale as old as time.
Underneath there is something darker, though. There always is. I refer to the juxtaposition of control and free will. Many ask why God allowed the Fall. They ruminate whether it was inevitable. Can AI be regulated effectively or will it slip through human fingers regardless of what is done? Is a middle way even viable? Do all paths lead to unexpected outcomes? Perhaps the true tragedy is that by the time we grasp AI’s full impact, our ability to steer it will be lost. This is the essence of the Collingridge dilemma, i.e. when regulating technology, the early stages are the easiest to implement change but there is insufficient knowledge about potential impacts. That is the “information problem”. Then comes the “power problem”, i.e. by the time the consequences of technology are understood, it has become so entrenched that control is expensive, time consuming, and difficult—if not impossible altogether.
Satan fell. Adam and Eve lost everything. The Spanish Inquisition arrived. Is humanity choosing wisely—or simply replaying an age-old tragedy, forever blind to the consequences waiting in the dark?
