Why AI Research Needs a Name Change
The distinguishing feature of the human species is no longer our ability to use tools, but rather our ability to create them. Two and a half million years ago our distant ancestors used stone tools in a manner that would ultimately separate them from other life on the planet(1). Today, our species can be defined by the invention, production, distribution, and improvement of tools both aesthetic and functional.
Where our technological progress was slow and incremental in the past, we now find ourselves rapidly improving and in some cases leapfrogging advances in science and medicine. 1941 CE marked the year the first fully programmable computer was created(2), only to be followed nine years later by an article written by Alan Turing proposing a cognitive framework can exist in a machine(3). John von Neumann, in 1955, began writing his lecture on the similarity between computer and brain that, though he was unable to deliver, was published two years after his death and is still a major influence in the computer technology field(4).
This paper does not assume any position regarding the likelihood that a computer will possess the cognitive capability of an average human or not. Instead, I am calling for a change to the name of a field that is actively pioneering this technology: artificial intelligence. Such a name change will better reflect our intentions and identify what we are trying to accomplish.
Artificiality is a word that accurately describes the current level of computer intelligence. By definition, artificial identifies an object as being human-made and an imitation of some other naturally occurring object(5). It is already incorrectly associated with a number of other concepts, such as artificial light. Light, for instance, is no more or less light simply because it originates from a bulb or other human-made source. Even artificial sweetener still possesses the qualities of being sweet even though it can only be produced inside of a laboratory.
Artificial limbs, however, offer an excellent contrast against the previous two examples and exemplify a proper use of the term. Prosthetic limbs do not possess the qualia vital to serve as functional replacements of a missing limb: they do not circulate blood, they are not ambulatory in the same manner as their natural counterparts, and they do not transmit tactile information to the brain. These limbs are certainly human-made and they are certainly imitations, thus denoting an appropriate use of “artificial.”
Right now a person can buy a phone that has artificial intelligence built into it. Talk-to-text software registers and computes a person’s voice pattern into text similar to how a stenographer records minutes of a meeting. There are even programs commercially available that can respond to basic queries about location and direction, weather reports, and other generally accessible information. A vast majority of the human population could provide the same services if you simply asked them and they possessed the appropriate information, so there is generally nothing special about what this technology can do. It is an imitation of the human intellect.
So, instead of defining a field in which we are aspiring to build beyond simple query-response tasks with terms that already signify our accomplishments, let us use another identifier. A term better suited to describing the level of computer intelligence we want to achieve is “synthetic.”
Synthesis in this sense is defined as a combination of two or more things to create something that does not occur naturally(6). Computer intelligence already satisfies the first condition; it is, or will be, a product of numerous materials assembled via chemical and electrical engineering and microtechnology with software to define the operational framework. The second condition is perhaps more difficult and even trickier to answer, though whether consciousness is in-and-of-itself naturally occurring is a metaphysical point. Ultimately, it is highly unlikely that silicon, copper, plastic and the other materials that constitute a computer will create a conscious experience without human intervention and this fact is what allows the second condition to be met.
The term synthetic also denotes the object to be something more than a mere imitation. Artificiality describes something to be an inadequate replacement, a mime, of another thing that already exists. By using synthetic as an identifier of what we want to accomplish in the field of computer intelligence, we maintain that the object (consciousness) is is not an imitation of intelligence, but rather a representation of a separate and distinct type of cognitive framework. Though our attempts at building a computer intelligence may be based in no small part on what we know of our own cognitive experience, the fact that such a cognitive experience exists outside of a human neurological framework in materials and code that is radically distinct from our own is sufficient enough to merit the term synthetic.
Synthetic intelligence can not only be used to signify programmed intelligence, but can also be an umbrella term for any cognitive system within a machine. Suppose we were able to copy your mind – every part of your conscious experience from your ability to conceive of ideas to your particular responses to stimuli – onto a hard drive. We then put the hard drive into a computer and provide the computer with perceptual instruments such as cameras, microphones, and even manipulators with which it could interact with the environment. Finally, we provide it with the software that enables your mind access over the system. Are you now an artificial intelligence or a natural intelligence in a machine?
The field of artificial intelligence has no answer to this, and yet by definition your copied mind would be a synthetic intelligence. It is a machine composed of human-made parts, assembled either by humans or machines, made intelligent by software that is not latent to the individual or assembled parts. Any intelligence in a machine is a synthesis of programming and engineering.
Innumerable examples of synthetic materials exist, most of which have trademarked names. Nylon and kevlar are two significant examples in that they redefined our application of woven materials and inspire further innovation of those materials and the invention of others. They are not found naturally and individually provide great benefits to our society. Nylon, for instance, was used heavily during WWII in parachutes(7). Kevlar is used in a variety of materials ranging from fiber optic cable to body armor(8).
Both of the above examples illustrate the significance of defining a substance as synthetic as opposed to artificial. Artificial is a term that borders on derogatory; an object that is labeled “artificial” is a substitute for those things that occur naturally. Yet nylon and kevlar are materials that are superior to materials found naturally in many ways. I do not intend to say here that computer intelligence is superior to human intelligence, but we certainly strive to make it, at minimum, equal to our own level of cognition.
My point here was simple: to provide an explanation of where our current descriptor of research into the development of computer intelligence falls short. Redefining the field as synthetic intelligence better explains the effort necessary to build a smarter computer, as well as displaying our intent to build machines that possess a cognitive capacity equivalent to a human’s. Our goal should not be to make a mimicry of our own cognitive systems and merely create thinking puppets. As scientists, we should, with all deference to ethical and moral constraints, push the envelope of computer programming to determine how capable of thought computers can be.
1) Choi, Charles Q. “Human Evolution: The Origin of Tool Use.” LiveScience.com. Live Science, 11 Nov. 2009. Web. 10 Apr. 2012. <http://www.livescience.com/7968-human-evolution-origin-tool.html>.
2) Hailey, David. “Technology History — Z3 Computer.” Interactive Media Research Laboratory at Utah State University. Utah State University. Web. 10 Apr. 2012. <http://imrl.usu.edu/OSLO/technology_writing/001_006.htm>.
3) Turing, Alan M. “Computing Machinery and Intelligence.” Mind Design II: Philosophy, Psychology, Artificial Intelligence. Ed. John Haugeland. 2nd ed. Cambridge, MA: MIT, 1997. 29-56. Print.
4) Von Neumann, John, Paul M. Churchland, and Patricia S. Churchland. The Computer and the Brain. 2nd ed. New Haven, CT: Yale University, 2000. Print.
5) “Artificial.” Definition for Artificial. Oxford Dictionaries. Web. 12 Apr. 2012. <http://oxforddictionaries.com/definition/artificial?region=us>.
6) “Synthetic.” Apache HTTP Server Test Page Powered by Linux. Princeton University. Web. 12 Apr. 2012. <http://wordnetweb.princeton.edu/perl/webwn?s=synthetic>.
7) Trossarelli, L. “The History of Nylon.” Caimateriali.org. Club Alpino Italiano, 2010. Web. 12 Apr. 2012. <http://www.caimateriali.org/index.php?id=32>.
8) “Uses and Applications.” Dupont.com. DuPont. Web. 12 Apr. 2012. <http://www2.dupont.com/personal-protection/en-us/viewall/uses_applications/50/14/uses_applications/USES%20AND%20APPLICATIONS/699.html>.