The singularity isn’t real, people

You, robot

Ray Kurzweil was a precocious child, to put it mildly.

His fascination with robotics and programming began early, and by the time he was a teenager he was already demonstrating the technological brilliance that would come to characterize his career. His pioneering approach to machine learning led to a television appearance at 17 where he demonstrated his new musical composition tool. Software he coded was included on computers sold by IBM. While in college he created a program that matched students to their ideal universities, which was sold for several hundred thousand dollars.1

His inveterate tinkering would later produce breakthroughs in areas as diverse as character recognition, optical scanning, and music synthesizers. Kurzweil founded multiple companies to capitalize on these inventions and went on to establish himself as one of the world’s most prominent advisors and speakers on technology and society.

But Kurzweil’s story wasn’t entirely without adversity. The early deaths of his father and grandfather impressed on him the frailty of the body and its inevitable decline. He responded to this genetic inheritance in proto-Silicon Valley hacker fashion, formulating a regimen involving a pharmacy’s worth of unregulated supplements and experimental treatments of dubious efficacy in an effort to stay young.2

This diligence might slow down the ravages of time but will not be enough to stave it off entirely, despite world-changing progress in other fields. In less than 70 years humanity went from flying a rudimentary airplane that barely left the ground to launching rockets that safely delivered astronauts to the moon. In an even shorter timeframe scientific discoveries led to everything from antibiotics to the decoding of the genetic scaffolding of life. Despite all this, aging and decline remain stubbornly universal.

Enter the singularity.

The concept itself is drawn from mathematics and the hard sciences, referring to a point where laws no longer apply and our understanding of certain phenomena breaks down.3 Kurzweil adapted the term to reference some vague future event when artificial intelligence crosses a line of such sophistication that it can then improve itself in exponential fashion, with consequences that are impossible to predict.

One of the more unlikely hopes emerging from this narrative is that computing power and human ingenuity will provide a way for people to augment their shortcomings, to the point where the constraints of biology no longer apply.4 At this stage disease can be forestalled such that lifespans become functionally unlimited. Brains themselves could be merged into some hypothetical cyber-consciousness, allowing unimaginable cognitive abilities and experiences.

Kurzweil wrote a famous book forecasting the impending arrival of this techno-rapture and has staked his hopes on living long enough to see it arrive. Major institutions support his Singularity University, which serves to popularize the concept across diverse fields. Deep-pocketed investors are spending billions on research programs which are expected to usher in this wild future. Venture capitalists can’t toss a pitch book without hitting something with artificial intelligence in the description.5

The law of accelerating returns is well on its way to upending everything we know. Right?

Category error

For several reasons, much of the hype around the singularity is plainly silly.

For instance, some people are already far more intelligent than the rest of us. Yet they don’t suddenly get smarter at becoming smarter, and thus transform through a runaway reaction into a transcendent intellect that eats the world.6 There is no reason to assume such a process exists for a machine. The fact that profoundly smart people have developed abstruse theories in support of the idea dissuades most from interrogating them too closely, but it doesn’t make them true.

Second, the logic that brains take inputs and make decisions, and so do computers, and therefore brains differ from iPhones in degree but not kind is simply not proven, despite what you might read in the credulous media. It’s claimed that if a program can fool someone into believing that it is a human, it is functionally indistinguishable from a person and might be given the corresponding dignities and rights. This is a little like saying that because a camera possesses a lens and looks at things, it should therefore see an ophthalmologist.

There is nothing magical about computational ability that somehow confers on it the properties of a self-aware, self-improving mind. Arrays of vacuum tubes have given way to transistors that have been miniaturized to an extreme level and packed into tiny chips, but these are in principle not that different from the devices they replaced, from the quipu to Napier’s bones. Presumably the early abacus factory had crates of finished products sitting around in a warehouse somewhere, and no one was concerned about this raw computing power suddenly usurping the human workers that created it.

One might respond that these tools lacked a power source and the ability to take in information through networks. Consider then the immense server farms overseen by giants like Amazon or Google, which are growing more powerful every day and have reached the point where our collective body of knowledge is readily accessible and the output of factories can quickly be sent anywhere. They continue to solve very specific problems, and only after the massive and continuous application of engineering intelligence.7 A calculator is not a mind, whether it fits on your wrist or takes up several data centers.8

The world is not enough

Paralleling the rise of singularitarian thinking is a growing theme of doomsday predictions, which are equally unsupported. Prominent executives have warned of an AI that could choose to enslave instead of serve our species. The singularity is supposed to somehow trigger the leap from bytes and circuits to the physical world through processes that are comically underdrawn, the Terminator series notwithstanding.

We have used technology to manipulate the real world for ages—a waterwheel is an application of intelligence that performs mechanized work on a scale that people could never hope to match, in the process likely putting many of them out of work.9 While retraining of manual laborers is needed in the short run, no one looks at these devices and fears for the future of the human race.

The nature of algorithms has become far too complex for an individual to understand, but the underlying principles don’t seem fundamentally new. Yes, these technologies may threaten to make radiologists redundant, but the basic issue is the perennial one of job dislocation and economic instability, not that our 21st-century equivalents of the slide rule might suddenly decide to go all Skynet on civilization.

Faster computers, better programming, and ingenuity advance the frontier of possibility. Barring societal catastrophe, they will continue to do so even while the line between visionary thinking and hype is blurred. But predicting that this will end in a singularity, or minds uploaded into an ecstatic hive brain, or something similarly wild, is a misguided response to our condition.

Predictions for when this singularity will arrive fluctuate but seem to stay just over the horizon—given that Kurzweil is in his seventies these timelines have taken on increased urgency. This supposed bridge to individual or collective immortality has precious little beyond hope to support it. This isn’t the first time that major social trends have decoupled from reality, and it won’t be the last.

It’s understandable that Kurzweil and others want to sublimate their existential dread into a quixotic research program, but will this mystically transform our lived experience into some techno-utopia? Not a chance.10 That’s in fact a good thing, because the hype obscures the real work to be done of understanding and then advancing the human condition, in ways that align with the real world.11

---

Not receiving these articles automatically? 1/week, subscribe here.

---

References

Microsoft co-founder Paul Allen deconstructs singularitarian thinking in the MIT Technology Review and Kurzweil responded.

Pando has another takedown of the concept.

Wired recounts Kurzweil’s attempts to achieve immortality.

Tim Urban’s epic longform blog Wait But Why has a characteristically overwhelming piece on Elon Musk’s Neuralink and the potential of AI. The whole thing seems a bit bonkers and probably is in critical ways that we’ll discover, hopefully before some test subject is left twitching in a cage.