HomeFeaturesExecutive Viewpoint 2017 Prediction: Sinequa – What’s Next for Machine Learning?

Executive Viewpoint 2017 Prediction: Sinequa – What’s Next for Machine Learning?

What’s Next for Machine Learning in 2017 as it Becomes a More Credible Workplace Tool?

There has been a lot of hype around machine learning for some time now, but in most cases it has yet to live up to the hype. But more and more frequently, organizations are learning how to bring together all the ingredients needed to leverage machine learning – and I think that’s the story for 2017. We’ll see machine learning move from a mystical, over-hyped holy grail, to more real-world, successful applications. Those who dismiss it as hocus-pocus will finally understand it’s real; those who distrust it will come to see its potential; and companies that apply ML to appropriate use cases will achieve real business benefit without the high cost of entry that was common in years past. In 2017 it will be clear that it has a credible place in the business toolkit.

In the past couple decades, we’ve seen hype around ideas like artificial intelligence and expert systems. The biggest difference between machine learning and this previous hype is that the four necessary enablers for machine learning – huge parallel processing resources, cheap storage, large and appropriate data sets, and accessible machine learning algorithms – are all now mainstream. Most large organizations have readily-available access to all these components (appropriate data sets are the only question, as they are always business- and use-case-specific), which makes machine learning a real possibility to reduce risk, increase customer satisfaction and loyalty, create new business models, identify patterns, and optimize complex systems.

One area where machine learning is growing rapidly and already showing success is for cognitive search and analytics applications; but it won’t take over core algorithms anytime soon. The quality and performance of specific search applications can be refined by applying ML to an enriched Logical Data Warehouse to deliver improved relevancy, identify relationships, classify content, and find and suggest similar material.  ML can be used to improve relevancy, but it is not a replacement for existing relevancy engines.

I don’t foresee machine learning achieving “mainstream” status in 2017, but it will within the next few years because the technology is advancing exponentially, quickly enabling its use in broader contexts.  We’re not completely out of the hype phase yet – many business leaders still need to learn how to separate legitimate applications from the (false) promises that ML is a panacea.  This distinction will become better understood in 2017 and as businesses successfully deploy ML projects with tangible benefits, the use of ML will quickly expand into broader areas.