When BI Becomes AI

1021

It’s hard to miss the fact that artificial intelligence (AI) has become the latest and greatest thing in the world of technology – and for good reason. AI is an umbrella term for a raft of technologies and techniques that allow software to behave in an intelligent manner. The word “intelligence’ is defined as an action that improves some measure of performance. In a sales situation that measure might be closed deals and/or value, or in a manufacturing environment the number of correctly identified flawed items.

So how is this related to business intelligence (BI)? Well, BI is a largely descriptive and diagnostic activity, where businesses simply aim to understand historical performance and/or why things have not turned out the way they should have – for better or for worse. To this end we produce reports and look at graphs and charts with the aim of gaining insights and understanding. It’s a manual, time-consuming, and expensive process, and as such ideal for the application of artificial intelligence. We can perhaps get a taste for how things might go by looking at BeyondCore, a BI platform that moves beyond descriptive and diagnostic analysis and includes automated analysis that highlights significant facts and hypotheses within the data. It addresses to some extent the serious problems associated with visual analytics – namely that interpretation is almost entirely subjective, often longwinded, and lacking in objective measures. It doesn’t completely do away with human intervention, but it does assist with the diagnostic and descriptive process.

If AI is starting to affect most areas of business activity – from selecting the best prospects in a sales situation, to robot delivery of fast food, we should not be too surprised to see it making headway in the world of BI. Certainly it is making a big impact on data mining and predictive analytics with several platforms providing largely automated creation of predictive models – PurePredictive being a good example. Applying AI to the process can speed up model development times by orders of magnitude – from months to hours.

BI will eventually succumb to the automation of data analysis through AI. An AI layer could at least learn from the users of a tool how the data is regularly manipulated, and to what end. However the real change will come when AI starts to understand our data better than we do. EBay recently acquired a business called Expertmaker, that does exactly that – extracting hidden value in data. In fact AI will add a new dimension in BI. Most of the time people are looking for what they know they don’t know. AI will power the discovery of information that users didn’t know that they needed to know.

There will of course be resistance – no one likes to think a machine can do a better job. But the application of AI to business intelligence would completely revolutionize its effectiveness and efficiency.