Tuesday, February 15, 2011

Innovation Requires Creative Intelligence

C. Engdahl
The Big E of Big E Toys

And, as imagination bodies forth
The forms of things unknown, the poet’s pen
Turns them to shapes, and gives to airy nothing
A local habitation and a name.

- Shakespeare, A Midsummer Night’s Dream

I’ve been a fan of the television game show Jeorpardy! for quite some time. And when I heard about the special competition (which began February 14) between the show’s two most celebrated champions – Ken Jennings and Brad Rutter – and an IBM computer named Watson, I was thoroughly intrigued.

A few years ago I spent a considerable amount of time investigating existing technologies and companies focused on understanding the vast array of consumer-oriented information found in the blogosphere. The company I was working for at the time was specifically interested in harvesting and understanding the “conversations” taking place in the blogosphere amongst consumers. We wanted to understand trends in consumer behavior and in some cases wanted to understand the implications of these conversations as they related to the specific brands of our clients - what were people talking about in the blogosphere?, who was doing the talking?, and were they speaking kindly or negatively about their subject? Some of the companies in question included Umbria, Scout Labs, Buzzback, and Collective Intellect. The software technologies themselves revolved around understanding “natural language.”

And it appears one of the great computer innovations in understanding “natural language” is Watson.

Enough has been written about Watson in recent weeks that I don’t feel obliged to actually talk much about it or describe its capabilities. I’ve watched it in action on Jeopardy! as well as seen the NOVA special and read other accounts of its development and exploits. To me it doesn’t really matter that the questions posed on Jeopardy! during the last few days have only seemed worthy of Jeopardy’s college episodes. Or that Watson has a distinct advantage over its human competition by being able to buzz in at precisely the right time. The machine is downright amazing.

Despite all its capabilities though, I can’t help think that everything about it and all the knowledge that it possesses was created by humans. Created being the operative would. Everything that Watson can do, and everything it knows, all the documents it searches and parses before offering a response was created by humans. Watson doesn’t really create anything of its own. The creative intelligence used to create Watson and the creative intelligence used to create all the information housed in its memory is even more amazing than the machine itself.

Innovation does not simply result from a set of rules or algorithms. It requires creative intelligence. Watson may be an innovation, but in its current incarnation, it can never be innovative.


Sean said...

I contest the notion that Watson has no creative intelligence. That is, in fact, the whole structure and purpose of Watson, to my understanding. A simple search routine is enough to find content references in the massive volumes of information in Watson's database, but the algorithms that work the magic are so much more than the serially-executed if-then-else of traditional computer school. The genius of Watson is that it has quantified the ~process~ of forming ~relationships~ across disparate data. Some definitions of creativity focus, quite explicitly, on the ability to combine multiple pieces of existing information in previously unrealized ways, and I have no doubt that there are humanly-unfathomable numbers of such relations that Watson has come up with which neither any human contributing the base knowledge, nor Watson's own creators, know or comprehend. If posed the right phrases- subjects for which, unlike in Jeopardy, no "right" response is known- I'm confident Watson could serve up responses based on those associations which would agree with or extend what a creative human might do with the same prompt.
To be fair, there are domains of human knowledge which Watson does not have and which it would be hard to represent to a computer- bodily-spatio-kinesthetic, emotional, musical, etc.- and types of associations which are currently just as intractable, and without those in the equation it will be hard to see a machine truly "thinking like" a human, but that's no reason to claim that the underlying processes are on the wrong track or incapable of showing creative results.

Vivek Menon said...

Did I misunderstand or were you just referring in your last paragraph about creating artificial systems to create content? I can understand the need to a Watson like system in today's world of a incomprehensibly mind boggling amount of information out there to make sense of it all. However if you asking for an artificial system which would create that content also - What would humans do?

C. Engdahl said...

Interesting comments. Sounds perhaps like the debate is simply about the meaning of "creative intelligence". Aside from this definition, the primary point in the post was simply that Watson's abilities (or if you prefer "intelligence") is based on systematic rules - albeit perhaps with some fuzzy logic built in, but rules and algorithms none the less. It's a completely rational entity. Innovation, and the requisite "creative intelligence" alluded to, isn't necessarily rational, isn't necessarily based on rules, isn't necessarily systematic. Where is the passion? The emotion? The gut feeling? The trepidation? The intuition? All are part of innovation, and a component in my mind of creative intelligence. Watson can perhaps mimic human creativity, but unless I don't fully comprehend Watson's processes (which is entirely possible), I don't think mimicking creative abilities is fully worthing of the innovation process.

Clicky Web Analytics