Comparing human and machine intelligence: philosophical approaches and criteria

Human and Machine Intelligence

The similarities and differences between human and machine intelligence doesn't seem to be the most important issue. It seems clear that both have been shown to exist, though they have very fundamentally different characteristics. The issue now centers more on supremacy: Is one better, more authoritative than the other? And if so, does this influence whether a "superintelligence" (Bostrom, 2003) exists that takes us to the paradigm when words (Zadeh, 2009) and emotions are most important (Dennett, Chapter 16)?

The early writings about projects like the Turing test tried to explain intelligence as being some kind of understanding about knowledge and its function. They often used simple conceptualizations similar to the way computers use the characters of "1" and "0" as a mathematical language. Philosophers use this approach to speculate about how a logical person might be able to "see" one color by itself, independent of another color that might actually be next to the first. The process of putting the colors together into a single vision became the focus of whether there was one or more persons in the understanding of what it means to be an intelligent human, and laid the ground work for looking at a few variables to understand intelligence (Parfit, Chapter 8). This approach allows for the argument about whether we and intelligence is about our "ego" (the experience we perceive) or a "bundle" of various experiences all tied together in several "persons" who exist at a given time. In either case, these understandings try to see knowledge, intention, meaning and such as identifiable separate entities that can be counted or measured.

Today's understandings of the differences between human and machine intelligence are based on more sophisticated views about the mechanics of computers and of the neurology of the human brain. One example of this idea can be seen in a series of posting on the Science Blogs. The writer, Chatham (2011), says that the first difference is the "Brains are analogue; computers are digital." Basically, this means that while brains send electric signals of messages around (in analog), computers use mathematical representations (using the digits of 1 and 0). These distinctions have important implications for what can be done with the knowledge. For example, the second difference notes how brains use "content-addressable" memory. The processor looks for the knowledge in question based on an association between the question and similar terms and concepts. Our brain "knows" that a fox is an animal or a pretty woman. Computers, however, are thought to just use "addresses," meaning they know where the knowledge is stored.

The remainder of this posting highlights other important distinctions between how the biological entity uses its knowledge verse how the computer hardware or software does. The last of the differences (that "Brains have bodies") is most relevant in that it gets to the ways in which human intelligence can interface with what is around it on many levels that computer hardware or software cannot. It should be noted, however, that this difference is not agreed upon by many experts (Block, Chapter 14). This is exactly why some see the interface capabilities as being the most important reason why we should be simply accepting that "superintelligence" is coming and get ourselves ready for the implications of the supremacy of machine intelligence (Bostrom, 2003).