Thursday, March 31, 2011

AI - BI - LI - "Watson - Come Here..."

Artificial Intelligence (AI), Business Intelligence (BI), and Location Intelligence (LI) -- how are they related? LI is a convergence of BI and Geographic Information Systems (GIS), and according to a recent article, we should expect a similar convergence of AI and BI in the near future. What will that convergence look like? And what will IBM's Watson have to do with it?

Bangalore-based Manthan Systems, a firm specializing in Retail Business Intelligence, believe the day of this convergence is approaching. But is this just another empty convergence promise? (I'm still waiting to do word processing on my toaster while I'm fixing breakfast.) Or are we actually moving toward "agile, intuitive and interactive solutions" that "will allow users to explore information freely rather than confining them to a predefined path of inquiry," as Manthan CEO Atul Jalan says in the company's press release?

Manthan's article asks the question: Can Watson Change Business Intelligence? Watson is, of course, the IBM-built computer that beat past champions on Jeopardy! You may wonder, why should we be amazed by a computer that can win a popular game show when other computers have been trouncing chess masters for decades?

The answer is heuristics. According to Wikipedia, heuristics "are strategies using readily accessible, though loosely applicable, information to control problem solving in human beings and machines." What this means, essentially, is that the human ability to solve problems is inextricably linked to our ability to learn and to apply experience. Traditionally, this ability has separated us from computers.

A computer plays chess using a body of human knowledge applied to precise movements within a limited scenario. Watson's capabilities need to span language comprehension, complex search of a large and broad range of data sources, and heuristic analysis to produce the most appropriate answers -- and to do it quickly.

Watson won at Jeopardy! not because of an out-of-the-box algorithm that gave it superior problem solving skills to its human competitors, but because its handlers "taught" it (or it "learned") how to play the game.

Is that real AI? Well, I don't think it would pass Alan Turing's test. But it may still have far-reaching effects on human-machine interactions. As the article points out, in the recent past, we've seen developments in search algorithms, mobile analytics, unstructured data analysis, predictive analytics, and collaborative BI. All of these developments point in the right direction.

I believe the greatest benefits of this gradual convergence and evolution will be "quality of insight" and "speed to insight," and the key to achieving better insights faster lies partly in the way the machine interacts with our queries, and partly in how it presents the various layers of applicable information for our consumption. Insight is not just about receiving information: you have to be able to grasp its significance and context as well.

Curiously, the Manthan article makes no mention of one area of BI convergence that is occurring now, and really is providing better quality of insight and speed to insight, and that is location intelligence. It's curious, because I can think of no market sector that will benefit more from LI than the Retail sector. Location Intelligence creates speed to insight by presenting complementary sets of intelligence (i.e., geospatial and textual) that allow the individual's very different left brain and right brain processing capabilities to work together, using whole-brain analytics to achieve a "Eureka" moment -- a flash of insight.

It's not AI, but it is a significant evolutionary step in the history of BI.

Judgment: the Manthan article is guilty of undocumented hype ("Experts say…"), but it makes a valid point about the evolution of BI as a more agile, intuitive and interactive medium.

*Alexander Graham Bell's first utterance into the newly invented telephone is reported to have been "Mr. Watson -- come here -- I want to see you."

No comments:

Post a Comment