This content is part of the Essential Guide: Guide to using advanced analytics and AI in business applications

Essential Guide

Browse Sections

Cognitive computing technologies still deliver mixed results

Cognitive computing tools have come a long way in the last couple of years, but the notion of true cognitive businesses, built around AI platforms, is still a long way off.

Despite substantial improvements in the utility of cognitive computing technologies, it's still too early for enterprises to go all in on the trend, and to base their business models entirely around artificial intelligence.

Of course, that doesn't mean some organizations aren't trying. At the Gartner Data & Analytics Summit 2017 in Grapevine, Texas, analyst Alexander Linden said, "We're seeing companies saying 'We are an AI company.' Some are even saying they're an AI-first company."

This surely delights IBM, whose marketing for its cognitive platform Watson (which it ambitiously claims can think like a human) focuses on convincing businesses the time is right to become a cognitive business.

Of course, any time a disruptive new technology comes on the scene, entrepreneurs and enterprising business executives will look for ways to exploit that technology. The Holy Grail is using new technology to create previously nonexistent business models.

Uber took advantage of the ubiquity of mobile data and GPS services to create a new way for people to find rides. That innovation earned the company a couple years of virtually competitor-free growth, and an undisputed lead over the competitors who have since come along. Everyone is now wondering what kind of new business models cognitive computing technologies will spawn.

Google might be one of the first companies to get there. It was nearly one year ago that CEO Sundar Pachai announced in a letter that the company would pursue an AI-first strategy. But, even with the advances in cognitive computing tools that we've seen in the 11 months since his letter was published, it doesn't make sense for the average enterprise to follow Google in this direction.

The reason is that any sort of general-purpose AI is still a long way off -- despite what some marketing campaigns might say. What we have right now are tools that can effectively execute single tasks. This means that, while it may be possible for enterprises to turn some specific jobs over to the algorithms, it's hard to imagine any business, beyond a large information-based company like Google, benefitting from it in a broad way.

That's one of the biggest takeaways from a new report out of MIT's Center for Information Systems Research. It says that narrowly defined tasks are the best use cases for today's cognitive computing tools. Good use cases today include things like banks evaluating customer creditworthiness, healthcare providers conducting insurance audits and accounting firms doing general audits.

But the report, "Five Things You Should Know About Cognitive Computing," states that any situation with high levels of uncertainty, rapid change or creative demands is not where AI thrives. The report recommends that enterprises proceed incrementally, applying cognitive computing technologies to areas where there are already strong business rules in place to guide algorithms, and large volumes of data to train the machines.

Not every business meets those conditions -- certainly, few businesses have those conditions across the enterprise. So chasing the notion of becoming a cognitive business makes little sense right now.

Next Steps

Cognitive computing may help users make sense of big data

Challenges of cognitive computing are more strategic than technical

Cognitive technologies demand the right use case, good management

Dig Deeper on Artificial intelligence platforms