Over the past month, I spent time at conferences dedicated to both poles of the analytics spectrum: business intelligence and artificial intelligence. I came away wondering why they're so far apart.
Both involve number crunching at their core. But while BI is primarily retrospective in nature, AI is all about the future. The statistical analysis behind BI is primarily about basic counts, in contrast with the ultra-sophisticated machine learning and deep learning algorithms that underpin AI.
Because of these differences, and the fact that AI as a useful tool is relatively new while BI is yesterday's news, we treat the two disciplines as wholly separate. During my time at the O'Reilly AI Conference in New York in late June, I heard not a single mention of BI. Conversely, at Dresner Advisory Services LLC's Real Business Intelligence conference held on the campus of MIT in July, AI was primarily used as a counterpoint to BI in discussions of the latter's strong business value and high adoption.
AI and BI: An idea whose time has come
But speaker Mico Yuk, CEO of consultancy BI Brainz in Atlanta, had an interesting idea. "I'm hoping that with more machine learning, key performance indicators will grow and evolve," she said. "I'm hoping we can work on metrics using data science and make KPIs work for you rather than you working for them."
In her vision, machine learning algorithms deployed in BI software will tell businesses what's interesting in their historical data. Right now, analysts typically have to define metrics for BI tools to track. It's a manual process and it only makes visible data the company already knows is important.
But machine learning-enabled BI could delve deeper into a business' unknown unknowns, finding insights in previously unexamined data. Taken a step further, AI-powered BI could take advantage of natural language generation capabilities to explain to the business what these insights mean and how they might act on them.
This isn't exactly a revolutionary idea. BI vendor Sisense announced in April a data discovery component to its software that automatically reviews data and alerts users to new and potentially interesting features. Other vendors are also adding machine learning components to their software. Analyst firm Gartner has been talking about smart data discovery since at least 2015. But it wasn't until this spring's Magic Quadrant report on BI and analytics platforms that Gartner declared smart data discovery the next disruptor in the BI software market.
Perceptions must change
There is some momentum around the idea of combining machine learning, business intelligence and artificial intelligence. But based on my reporting, the idea seems to have seeped little into the general public's consciousness. We still mostly see AI and BI as separate domains.
This has to change. Think of all the time AI-powered BI platforms could free up for analysts, who currently spend much of their time handling requests for reports. They could move on to more effective data science and predictive analytics projects. We've seen from companies like Uber and Google that smart approaches to machine learning can create entirely new, profitable business models. That's the kind of thing people with strong analytics skills should be spending their time on.
Maybe once the hype surrounding AI starts to wear off the differences between it and BI will seem smaller and the opportunity more obvious. After all, we're already seeing AI having a big impact on CRM, HR and FinTech software. There's no reason BI shouldn't be next.
Tools we call AI today rarely live up to their hype
Sharp developers needed for building strong AI tools
Why AI might be for real this time after many false starts