tbeltrans
Senior Member
In CompSci circles, pure randomization is not considered possible. However, randomization within certain parameters can be applied to various generative algorithms to create the effect of what we might consider creativity. I suspect that as these areas continue to develop, there will be (if not in some areas already) a time when most people will not be able to discern the difference between human and machine "creativity".
In music for example, consider jazz improvisation. Though it can sound endlessly creative, the reality is that each musician has his or her own particular vocabulary of practiced licks that s/he pieces together in various ways, just as we all typically use a vocabulary of somewhere in the vicinity of 200 words in daily conversation. These practices can be recreated with a well designed randomization algorithm applied to a specific area of computer "creativity" because in human application, they are similarly limited to begin with.
This is not some idle conjecture, but instead based on my own practical and learning experiences in each of the areas I mentioned. Computer chess is different in that there are programmed book openings and end games, with the middle game largely dictated by a tree search across all the possibilities for those stages of the game. Some chess software has attempted the use of learning algorithms to "learn" a player's style and adjust to suit that, but most still use the materials I mentioned earlier.
Tony
In music for example, consider jazz improvisation. Though it can sound endlessly creative, the reality is that each musician has his or her own particular vocabulary of practiced licks that s/he pieces together in various ways, just as we all typically use a vocabulary of somewhere in the vicinity of 200 words in daily conversation. These practices can be recreated with a well designed randomization algorithm applied to a specific area of computer "creativity" because in human application, they are similarly limited to begin with.
This is not some idle conjecture, but instead based on my own practical and learning experiences in each of the areas I mentioned. Computer chess is different in that there are programmed book openings and end games, with the middle game largely dictated by a tree search across all the possibilities for those stages of the game. Some chess software has attempted the use of learning algorithms to "learn" a player's style and adjust to suit that, but most still use the materials I mentioned earlier.
Tony