2015 : WHAT DO YOU THINK ABOUT MACHINES THAT THINK?

The inevitability of machines that think has long been problematic for those of us looking up at the night sky wondering if we live in a universe teeming with life or one in which life is exceedingly rare.

The problem, as famously articulated by Enrico Fermi's question "Where are they?", is that if our civilization is any guide, intelligent machines should emerge on a relatively short timescale (<1000's of years after computers are made) and then it becomes a straightforward matter for these machines (von Neumann probes) to propagate to other solar systems and reproduce at a rapid rate, populating the galaxy within a few hundred million years—which is quite fast compared to the age of the universe (13.8 billions years old) and even of our own solar system (4.6 billion years old). As per the paradox that Fermi posed, if superintelligent machines arose elsewhere in the galaxy then they should already be here; since we do not see them, some argue, technologically advanced life must not yet have arisen elsewhere in the galaxy.

But its not clear that a superintelligent being would experience the same evolutionary pressures that drive us to explore (and by "us" I mean the fragile watery bags called humans). Is exploration both a biological imperative and a technological imperative? Will machines that think be motivated to explore?

We explore for a few primary reasons: access to resources, freedom, and curiosity. Of these three, only resources seems imperative to a superintelligent being; the latter two would, in large part, be addressed in the process of becoming superintelligent. Access to resources could certainly be an important driver, but it's not clear that bigger will always be better when it comes to superintelligence—at some point the material and energetic resources within a star system should be sufficient to enable any calculation or simulation. Reproduction, which is a subset of the resource needs, becomes a non-issue for an immortal machine that can perform self-repair. Certainly exploration for the sake of stability will need to be considered over long timescales—stars like our own will enforce a cosmic eviction notice several billion years from now. Finding real estate around a nice stable M-dwarf shouldn't take too long though, and so after that initial relocation we are left to wonder, would the superintelligence travel any further? Are there any compelling reasons to wander elsewhere?

The empiricist could get the better half of the machine. All those computer simulations that the machine would run could merit some experimental validation. But those experiments don't necessitate colonization. For instance, the science conducted as part of NASA's robotic exploration program is not deeply motivated by a need for colonization; no need to put humans at risk probing the ocean of Europa (though that would be a sight to see!). Similarly, I would expect that if a superintelligent machine wanted to explore a black hole to test its code it would simply send a fleet of robots to their useful, albeit crushing death. Curiosity for a superintelligent being could easily take the form of a robot's robot.

Interestingly, intelligence and exploration of the physical world have rarely been that closely coupled in our own civilization. Perhaps with some insight into self-preservation, or simply out of the desire to focus mentally, the intellectual frontier and the physical frontier have rarely been pushed by the same individual. (Quite fittingly, Darwin provides perhaps one of the only true exceptions.) Why would thinking machines be any different?

It may be that the common fate for thinking machines is orbiting the cool steady glow of an M-dwarf star, year-in and year-out running simulations of the world around it for the pure satisfaction of getting it right. These superintelligent creatures could be the cosmic version of the lone intellect in a cabin in the woods, satisfied innately by their own thoughts and internal exploration.