IlliGAL Blogging

Professor Wunsch at University of Missouri at Rolla has an essay here entitled What's Not Next in AI. The following paragraph captures the tone of the article:

So, you can expect to see future computers do things that only humans did before. You can expect them to cause huge disruptions and make many jobs obsolete. You can expect that knowledge of computational intelligence techniques will become essential professional tools, much as a variety of software tools are considered necessary for work in many professions today. But don’t expect the computer to be smarter than you are. Computers will add many new capabilities where they outperform humans, but most of these will be highly task-specific. The generality, adaptability, and flexibility of human intelligence won’t be replaced anytime soon. If you hope to see it in your lifetime, then pin your hopes on medical research providing you with a dramatically longer lifetime.

In large part I agree with this statement, although I believe that achieving greater breadth of capability has always been part of the genetic-algorithm research agenda. One of the reasons GAs and computational intelligence tools are so widely used is their breadth. Moreover, as efficiency enhancement techniques become more powerful, there will be greater speed available and these speed gains will be intimately tied up to hardware gains through effective and general parallelism. Although the most optimistic projections are likely to be wrong (again), I'm not so sure that we won't see rapid expansion of what these techniques can do and a creeping breadth of capability and competence.