Gartner: Why there is No Intelligence in AI

AI is not one thing, it's many. There are many, different amazing innovations at various levels of gestation. I'll hazard a guess that, in the Gartner Hype Cycle methodology, we will not have a single point labelled "AI." We'll have many points in an overall AI hype cycle.

We have in the industry many vendors trying to market their single AI entity, as in Cortana, Watson, Holmes, Alexa, Amy and so on, as though they were intelligent entities, like your next-door neighbour Albert.

Don't buy the metaphor. It's not a single entity and it's not a person and it doesn't function internally the way a human does, no matter what its name is (my 1962 Morgan +4 Roadster has a name — MOGGIE — that doesn't mean it's a person either.)

There is no magician behind the curtain. There are many different algorithms woven together along with massive amounts of data to produce a capability that appears Amazing, at least for now.

It is also an illusion — a deception perhaps — to call any of this intelligence. Trace back the history of IQ, its role in the eugenics and immigration debates of a hundred years ago and look carefully at the scores of academic definitions of intelligence (there are many.) We don't know what human intelligence is and we suspect it's a measure of cultural knowledge, not the ability to act.

"Intelligence" is a marketing concept. When we call a technology "intelligent" we are implying that it has some mysterious quality akin to a human's (or even a mouse's) intelligence (whatever that is.) Ridiculous. Airplanes don't have the native abilities of pigeons but they can fly like one! 

Then there's the marketing term some others are using, "Augmented Intelligence" to emphasize the notion (which I endorse) that AI can help make people more effective at what they do best.

But Augmented Intelligence is a term too many use for political correctness and to avoid the fact that they will accept orders to use their technology to put people out of work too.

ACTION: Step back and see the Amazing Innovations now occurring in deep neural networks and derivative natural language processing tools. Watch how they are applied to enable new business results.

And plan carefully: today's greatest innovations in those fields will "soon" be obsoleted by even more amazing development. (Gotten bored with all the heretofore Amazing Innovations in your smart phone yet?)

Get amazed! Exploit. Prepare for rapid and disruptive evolution in AI. And don't forget technologies that Formerly Were Amazing (FWA). Many of them may be a better bet, in some cases, than the truly amazing stuff of today.