h+ Magazine

Imagine discovering a town packed with immobile cars, some crashed into each other, others parked at odd angles as though their occupants just abandoned them. There are no people anywhere. The drivers and passengers have fled the scene. Only, they did not flee on foot. A greater power has taken them to a better place. They were the chosen few who have been saved by the Rapture.

Jaron Lanier would recognize this scenario, because he has seen it. Not for real, but in many paintings hung in gas stations and other mundane places in New Mexico, where he grew up. To him, this American evangelical version of the Christian apocalypse is directly comparable to the concept of a technological singularity. In You Are Not A Gadget, Lanier informs us that “singularity books are as common in a computer science department as Rapture images are in an evangelical store.” And, in case that has not sufficiently established the connection, he later tells his readers what the singularity involves: “…people dying in the flesh and being uploaded into a computer and remaining conscious, or people simply being annihilated in an imperceptible instant before a new super-consciousness takes over the Earth.”

People who write about the singularity sometimes discuss such possibilities as well. Ray Kurzweil believes we may one day upload ourselves; Hugo de Garis thinks artilects (artificial super-intelligences) might decide to wipe out the human race. Lanier is right in saying Rapture-style scenarios are anticipated by some folks in computer science, but I think he is wrong to say such scenarios define the singularity.

So, what is the correct definition? Simply this: The creation, by technology, of greater-than-human intelligence. This is often thought to involve computers developing artificial super-intelligence and leaving humans far behind, cognitively speaking. But that is just one way in which Vernor Vinge — generally accepted as the originator of the concept — imagined a singularity might be brought about. In his 1999 paper, he outlined several scenarios, most of which involve an increasing symbiosis between people and technology.

If a singularity ever happens, it would have two future outcomes. It would make tomorrow “fast” and “strange.” The physicist Michio Kaku divides “impossibilities” into three categories, which he calls types I, II and III. Basically, type I is anything that either has a rough working prototype, or at least some basic practical proof-of-principle. Type II refers to possibilities on the very edge of our understanding, concepts likely to remain entirely theoretical for many generations to come. Type III are those things which are ruled out by the known laws of physics.

The “fast” effect is usually described in terms of a shrinking timeframe between successive generations of artilects. We succeed in developing an artificial ultra-intelligence. Being able to do anything we can do, this machine designs a version 2 that is superior to itself. Thanks to its more advanced brain, it achieves this goal sooner than we took to design version 1. Then version 2 designs version 3 sooner than version 1 took to design version 2, and so on. Artilects that mere humans would have taken a million years to design (if ever) are being designed in decades… years… weeks. Furthermore, these vastly intelligent entities take type III impossibilities and figure out how to accomplish such things in no time at all. Again, capabilities we did not expect to be realized in a million years are happening much sooner.

“Strange” supposes these artilects will develop models of reality that replace our own. We know our theories are incomplete, and so we cannot be absolutely sure that everything we categorize as a type III impossibility really is impossible. Who knows, maybe posthumans will develop a quantum theory of gravity using conceptual models beyond human understanding and create technologies dismissed as impossible, or never imagined by humans at all?

I don’t know if we can succeed in creating a singularity or not. I don’t dismiss the possibility, nor do I subscribe to it as an inevitable future event. But I do know that it need not happen “soon.” And it need not involve anything like the uploading or the annihilation of the human race. Critics often describe a singularity as being a near-term event; something its believers expect to happen years from now, or a few decades. Some thinkers do anticipate a singularity within such timeframes, but it does not have to happen within such timeframes in order to be valid. We might find that the world in 2065 is nowhere near as “fast” and “strange” as expected, despite Kurzweil’s prediction that the singularity will occur around 2045. But that does not mean we can conclude, “oh well, the singularity must be a load of nonsense.” Profound transformations in human capability might occur centuries or millennia from now. No matter. It would still be reasonable to define the transformation as a “singularity.”

As for “apocalyptic” scenarios, you do not have to take such things seriously in order to credit the singularity concept with some degree of truth. Ideally, comparisons of technological advances with Rapture-style images would be discussed using some other term — perhaps we should use the term “Technocalypse” (coined by Frank Theys). The singularity, meanwhile, would be debated in ways more comparable to aviation in the pre-flight era. Some eminent scientists (most famously, Lord Kelvin) dismissed heavier-than-air flying machines as an impossibility. Similarly, some modern scientists doubt that cyborg implants will ever surpass natural senses; or that wearable devices will ever amount to a global supermind; or that genetic engineering will ever lead to a race of elite superhumans. In contrast to aviation, we currently lack the 20/20 hindsight that allows us to dismiss such doubts about singularity. For all I know, all doubts may be valid.

But our predecessors could be doubtful of flying machines without directly comparing its pioneers with religious fanatics. Similarly, we can and should debate the scenarios Vinge outlined as possibly leading to a singularity without comparing its pioneers with evangelicals convinced that a Rapture is looming on the horizon. Yes, there are individuals working in robotics or biotechnology who harbor Rapture-style hopes for their work, but if you think about it, given that a singularity represents a conceptual barrier to further understanding, it hardly seems appropriate to tie ancient prophecies to the concept. How can something anticipated by our ancestors represent concepts triggered by current developments? Mind uploading, Matrioshka brains and other such concepts are weird to some, but not nearly weird enough to qualify as a product of a singularity. So what would qualify? Ah, that is the problem. You would not understand, and I could not tell you.

If you cannot stare into the singularity and report back what you see to “mere” humans, who can you communicate with? Presumably, entities capable of grasping concepts that cannot be framed in any human language. How interesting, then, that Lanier’s book (which begins with a dismissal of the singularity as Rapture-style nonsense) should end with a brief discussion about “postsymbolic communication,” which Lanier describes as an “extraordinary transformation that people might someday experience.” What if a radical new form of communication, unlike anything we have today, evolves in the future? Surely, then, our descendents could describe dreams to each other that we cannot imagine? And if they act on those dreams and invent new technologies, they, too, would lie beyond our ability to imagine. So, you see, Jaron Lanier does believe in the technological singularity after all.