In the modern world, human cognition is supplemented by a vast array of distributed and highly specialized artificial intelligences. Each performs a specific function, often with superhuman ability, and their combined efforts allow individuals to extend the breadth and scope of their interactions and influence. But to what extent is human action (and even thought) supplanted by machines? In Me++, William Mitchell suggests that interactive computing has gone far beyond the restricted arena it was conceived in. Machines are no longer distinct, individual mechanisms, architectural in scope. Instead, through a network of miniature devices, computing has become integrated into almost every aspect of our daily lives.
My local stock of neurons has (the neuroscientists gloomily assure me) been diminishing as I grow older, but the supply of silicon and software at my disposal has been growing rapidly. Consequently, the neural network inside my cranium oustources more and more mental functions. I don't do much mental arithmetic any moree; calculators and computers take care of that. I don't rack my brain for half-remembered facts; I look them up on the Web. I routinely exist in the condition that J. C. R. Licklider presciently identified, way back in 1960, as "man-computer symbiosis" -- except that Licklider, Doug Engelbart, Ivan Sutherland, and other pioneers of interactive computing mostly had dialogue with desktop workstations in mind, whereas I now interact with sensate, intelligent, interconnected devices scattered throughout my environment. [Mitchell, 34]
Are we replacing an important part of ourselves with technology?
Our ability to function as a society relies on networks of electronic devices that might fail. Our technology could be disabled by a disaster or catastrophe that might not affect biological organisms. Could we regain our lost individual, personal functionality that had been outsourced to computers and other gadgets?
Is there cause for concern that our memories (and other faculties) may suffer as a result of our increasingly dependent relationship with computers?
As technology extends our capabilities and increases efficiency, it is also gradually phasing out human involvement in our own lives -- this is troubling.
I don't directly control all the functions of the machines and devices I use; I rely on the intermediating machine intelligence embedded in my cellphone, my car, my domestic appliances, the operating system of my laptop computer, and my software agents. Mostly I cannot tell whether such intelligence is supplied by local devices, by remote servers, or by some combination of the two, and it doesn't matter -- as long as there is capacity available somewhere, and the connections are sufficiently fast. Often I cannot even tell whether a verbal response I receive over the network has been generated by a person or by a machine; the Turing test has stealthily been aced. And when I'm in a deathmatch on a first-person shooter like Quake, an AI opponent such as ReaperBot may be more formidable than a human one. As nodes of machine intelligence are distributed just about everywhere, as electronic interconnectivity grows, and as electronic feedback loops multiply, cities are evolving into extended minds and biological brains are becoming elements of larger cognitive systems. It is Santa Cruz guru Gregory Bateson's "ecology of mind," but with much more silicon and many more electronic interconnections than he ever imagined. [Mitchell, 35]
Controlling an AI opponent in a first-person shooter is a good example of an algorithm well-suited to computer intelligence. What tasks are more difficult for machines to learn and perform? Understanding human language and 'common sense' still seem to pose a problem for computer programs. Have machines truly passed the Turing test in the area of natural language processing?
In my experience, while voice synthesis of any given phrase may be nearly indistinguishable from human speech, verbal interactions with AI are still not easily mistaken for interpersonal conversation. Considerable progress towards machine fluency has been made through computational linguistics -- particularly statistical natural language processing -- but the work is far from complete.
Does the complexity of interconnections and multiplicity of electronic feedback loops create a chaotic world, in direct opposition to the order of the natural world? Although technology and nature are to some extent mutually exclusive, we have modeled aspects of our innovative electronic networks and other technologies after processes and relationships observed in nature.
Mitchell, William J. Me++: The Cyborg Self and the Networked City. Cambridge: MIT Press, 2003.
Last modified 7 February 2005