Singularity: Nanotech or AI?

Engines of Creation by Eric Drexler. Photo: nano-scientist.orgThe question of the relative roles of nanotechnology and AI in forging the shape of the future has been argued in techno-futurist circles for decades. Eric Drexler mentioned AI as a potentially disruptive technology in his seminal 1986 book Engines of Creation, and it was discussed at the very first Foresight conference 20 years ago.

It is generally assumed that a self-improving super-human level of AI is part and parcel of the Singularity, and indeed, such was the basis of I. J. Good’s and Vernor Vinge’s conception of the “intelligence explosion.” But let’s assume, for the sake of a scenario, that creating self-improving AI is just a lot harder than we think, and that we aren’t going to invent it until well after we have flat-out molecular nanotech with the ability to build fast self-replicating diamondoid nanomachines. What then?

One thing Drexler predicted in Engines was that without needing to create true human-level intelligence, automated design systems — narrow as opposed to general AI — would enable the creation of highly complex nanosystems, well beyond the capabilities of mere human designers. How did that prediction pan out? I would have to say that it was so accurate, and happened so soon, that it’s taken for granted today — human designers with only pencil and paper would have no chance of designing, say, a modern computer, or indeed any of today’s complex engineered systems. Like many areas, design automation is an area that was once considered AI, but isn’t any more.

What does a Singularity look like with just nanotech and narrow AI? Let’s consider the standard list of transhumanist concerns:

LIFE EXTENSION: Playing around with the interiors of our cells and so forth is clearly a nanotech application. Uploading or radical body improvement is the same.

AI: Ultimately, we get AI by uploading, doing lots of neuroscience, and understanding how the brain works. We get human-level AI but not super-intelligent ones. We do ultimately get faster ones — but our uploads can be faster too.

PERSONAL NANOFACTORIES AND UBIQUITOUS WEALTH: Nanofactories wouldn’t be quite as powerful without a superintelligence to drive them. They could only make what someone invented and designed, rather than inventing things themselves. But that would be enough to kick the entire physical economy over into a Moore’s Law-like growth mode, eradicating hunger and poverty in a decade or two.

FLYING CARS, SPACE TRAVEL, OCEAN AND SPACE COLONIZATION: Again, these are clearly nanotech applications. The modifications to the standard human body necessary to thrive in space require significant nanotech capabilities.

ROBOTS: Robots with human mental capabilities and virtually any physical capabilities would be straightforward, and would rapidly become affordable for everyone.

All of these areas require more scientific knowledge than we have now, but not more than the current rate of scientific progress human scientists are likely to produce in the next few decades. The current techniques of narrow AI are capable of automating pretty much any well-defined task, albeit with more programming effort than would be necessary if the machine could learn for itself.

The modifications to the standard human body necessary to thrive in space require significant nanotech capabilities.

With its Moore’s Law rates of increasing capability and reducing costs for really high-tech physical equipment, one of the things that a nanotech revolution could do is to make scientific instrumentation ever more available. Existing efforts toward open-source science would be enhanced and given more headroom. The scientific knowledge, ingenuity, and experience necessary for the full utilization of the physical capabilities of nanotech could grow as rapidly as the Internet and cell phone use has over the past couple of decades.

So, back to the present. I don’t really expect AI to lag behind nanotech as much as this analysis suggests. In fact, I think it will precede it. But even if AI were to stall at roughly its current level of capability, something like a Singularity, a reprise of the Industrial Revolution that boosts our civilization from terrestrial to solar, making us all long-lived, healthy, wealthy, and maybe a little bit wiser, is not only possible but very likely.

J. Storrs (“Josh”) Hall, PhD., is president of the Foresight Institute, founding Chief Scientist of Nanorex, and author of Beyond AI and Nanofuture.


Leave a Reply