"Once more, folks are confusing AI with ultrafast computation, pattern matching and huge database searching, techniques that are easily done by digital computers and result in success at algorithmic or database searching 'games' where digital computers outperform human capabilities."

On the contrary! Much of why the architecture of the human brain stops the spit out of some of our best attempts is the raw power of billions of surprisingly sophisticated parallel processors arrayed in remarkably structured neural networks to do just these things: Pattern Matching (in visual systems, for example), ultrafast computation, and 'searching databases' (say, for example, in memory! Which the human brain is amazing at!).

"The fundamental mistake is in defining intelligence as possesing such abilities, admitedly understandable because these attributes are uniquely displayed by the human species. "

Oh, really? Would you define any system as intelligent if it did not have something close to human like memory ('database look-ups'), something close to human like ability in perceptual systems (pattern matching), 'problem cuing' (pattern matching), analogical reasoning (pattern matching), speech recognition (ultrafast computing)... etc etc etc?

Oh, and while you are pontificating- why don't you advance a solid notion of what you mean by 'Intelligence'? I'm just dying to hear it.

" Real world interaction with an environment is NOT readily done by digital computers and yet is easily done even by very simple life forms. "

Indeed, thats because the biomechanical, nervous system, system integration and even synaptic processing power of relatively simple biological entities stomps the stuffing out of all but a few of our analogous experiments. We make pretty decent 'synthetic' roaches and other insects, though, last I checked. Some that even very well exhibit some of the things claimed to be impossible in your referenced critique.

"When one makes factual points with examples that several decades of AI predictions have fallen short and still continue to do so, the rebuttal is indignation, ad hominem accusations of religious fervor, and vague hand waving about some indeterminate future."

My question to you is whose predictions? And shall I remind you, that science in general does not include the generation of timelines by which we say we'll have certain advances. That is purely a speculative and non-scientific endeavor. Shall I fault Physics for not having Fusion '20 years form' 1950? Science sometimes blows the doors off timeline speculations, sometimes it doesn't. The variable isn't the 'suckyness' of the science or the failures of the field- the variable is that you don't know how hard a problem is until you solve it!!!. Thats why timeline statements are not science and are entirely speculative endeavors, regardless who utters them- be it scientist or layman.

Indeed, many of the HUGE goals placed out there for AI have been met: We have systems that can far outperform humans at things like chess. We have perceptual systems that can drive cars over rough terrain and weave in and out of real-world traffic. We have Artificial Life in the form of mechanical insects. We have Automated Theorem Provers (and have since the 50's!) we have network based classifiers. We have systems that can fly planes. We even have systems that can pass pretty sophisticated turing test environments.

It just turned out that our original formalizations of what we mean when we say 'intelligent' have changed, matured an become deeper (despite the bloviations of Mead). We now know that intelligence is a lot more than doing any of the above things.

Yeti