2015 : WHAT DO YOU THINK ABOUT MACHINES THAT THINK? [1]

noga_arikha's picture [5]
Associate Fellow, Warburg Insitute (London); Research Associate, Institut Jean Nicod of the Ecole Normale Supérieure (Paris); Author, The Ceiling Outside
Metarepresentation

 

The history of humanity and the history of technology are conjoined. We have always used our cognitive capacities to create the objects we needed to survive, from weapons to garments and shelters. The evolution of the human mind is instantiated in the evolution of technology. We have developed a capacity for metarepresentation—a capacity to be aware of having, and to analyze our own minds—which is a function of higher order consciousness. And in order to look at ourselves in the mirror, we have always used technological analogies, compared our minds to the technologies we had created. To each era its machine—from hydraulic pumps to computers.

We have by now created technologies that no single person is able to master. Our creations are starting to escape our own minds. No wonder then that we so easily imagine the creations becoming creatures in their own right, endowed with minds as agile as ours, or more agile perhaps. Science fiction imagines perfect robots, indistinguishable from ourselves, embodied, speaking, seemingly feeling, that can fool and even perhaps attack us.

But in thinking conceptually about our own minds, we tend to remain Cartesian dualists. Thinking seems so disembodied an activity that we forget that we are emphatically not brains in vats, that no amount of microtechnology will recreate the complexities of biology thanks to which our brains function, replete with neurotransmitters, enzymes, and hormones. We are our bodies, we have emotions that are embodied and that deeply inform our thinking processes. Machines are developing task-driven cognitive capacities, but their perfect processing is very different indeed from the imperfect, inconstant, subtle thinking of persons endowed with a sense of self, proprioception, a sense of centeredness, the "qualia" that distinguishes us from "zombies."

Computers excel at processing processes most of us fumble with, and we are increasingly accessing the world of facts via machines. Much of our memory is assigned to Google, and there is no doubt that our minds are increasingly extended beyond our single bodies, that we exist within an increasingly large network of disembodied minds and data. Thinking is itself in part a socially given capacity, and to think is to participate of a collective enterprise. But the complexity of this enterprise is as much a characteristic of the human condition as is our embodiment. Machines do not have social lives any more than they are embodied within a complex, evolved set of biological tissues. They are good at tasks, and we have become very good at using them for our purposes, and for expanding our capacity for communication. But until we replicate the embodied emotional being—a feat I don't believe we can achieve—our machines will continue to serve as occasional analogies for thought, and to evolve according to our needs.