(Up to: Technology Vs Evolution )

There are some things in computer science that appear to some to be mere science fiction, or futuristic pipe dreams, especially in the fast-moving, instantaneous environment that the rest of the industry has generated. Speech recognition, or more specifically interpretation, is one of these - the ability for a man-made machine to understand what a human means through words.

Attempts so far to understand the semantics of natural language are generally limited to a small, specific domain of words and knowledge. Programming languages are a detailed, formal version of this. The other end of the scale is far from being achieved - getting code to the level at which it can not only relationally link words to other words, but also to objects, emotions, experiences. Personally, I think that this level of comprehension is implausible in a purely software domain. Without sensors/senses such as we have, a program is unable to truly associate language elements with what it experiences. However, it may be possible to derive a system of "machine comprehension" that is based upon these links, but without direct access to the senses. i.e. perhaps it is possible to create an emotional system that has learned via access to sensors, but can be separated from them as solely the "processing" "unit".

This "automation" of information - the codable aspect to semantics - is a chilling prospect. Once it has been achieved, once language can be classified, comprehended, created, sorted and manipulated by something that we have created, then we begin to trespass yet further on the domain controlled by human thought. We all have the ability to tell what is "good quality" and what is "bad", in terms of creative works (although the conclusions we each draw are obviously subjective). Should this become "computerisable" (in the loosest sense of the word "computer"), then we can effectively reduce the artistic process to that of a script. We are able to draw up new "talent" simply through the mass-production of material, and then filtering it according to some automated selection process. Similarly, we can link material to other material more fully than ever before - this is starting to happen now, but has a long way to go before losing its formality and structuredness that it necessitates currently.

Imagine a web where any possible material has been generated, ranked and linked, an electronic typewriters and monkeys scenario, at least in terms of creative subjects. (The ability to tell truth from falsehood and to evaluate empiric research is a different matter.) We would do away with our strive for originality, and the only reason for us to create such works ourselves would be purely for individual satisfaction. Perhaps that is good, perhaps not so.