For tech giants like Amazon and Google, machine learning has long been central to their operations - the most famous example being Amazon’s recommendation engine, which many see as having been the key to its success. Such technology has huge implications for organizations across all industries though, and the wealth of data that companies now hold, along with the rise in affordable products like Microsoft Azure ML and IBM Watson, mean that they are rushing to adopt it.
However, along with any technology, there must be an element of caution exercised in adoption. David Linthicum recently wrote on inforworld.com that, ‘Machine learning is valuable only for use cases that benefit from dynamic learning - and there are not many of those.’
He continued, ‘The problem is if you have a hammer, everything looks like a nail. Vendors pushing machine learning cloud services say it's a good fit for many applications that shouldn't use it at all. As a result, the technology will be over-applied and misused, wasting enterprise resources.’
Linthicum is correct to say that vendors’ sales pitches should be treated with skepticism, but when he says that there are not many use cases that benefit from dynamic learning he seems to be going too far in urging caution. Machine learning is an extraordinary tool when it comes to analyzing any amount of data alongside every combination of variables, and there are a multitude of use cases out there.
Clearly, as Google has shown, it’s essential for analyzing users, and the insights it provides can deliver highly-personalized content and ads - far more so than the type of predictive analytics tools that are now commonplace. Other firms are now realizing the potential in similar ways. Home Depot, for example, uses it to find goods from its large inventory, such as bathtubs, and connect them with customers’ specifications.
Perhaps the most profound consequence of machine learning for businesses is in decision making. Analytics has long been used by decision makers as evidence, but only by answering the questions that they think to ask of the data. Machine learning should push past this and effectively automate the decision-making process. In such a world, systems would be proactively informing you about what might happen and what you can do about it.
Machine learning algorithms can also act as a personal assistant. So when it looks like you're going to miss a deadline, it can reschedule everything that relies on that deadline being met. For example, if a website is not going to be complete until three days past the deadline, it can recommend you postpone posting promotional materials linking to the site until it is up, all without being prompted.
Essentially, machine learning makes your organization far more proactive than it currently is, and automates tasks in such a way that can greatly reduce costs and wasted human effort. According to Gartner, almost every business unit is going to be interested in these tools. Results have already been seen, with papers showing many cutting costs by up to 70%, and revenue going up 20 times because of increased speed in tracking buyer behavior, and improved customer service.
The one drawback at the moment is people’s comfort with automation and turning things over to machines. Ultimately, this will change as the results are realized and automation becomes more of a factor in our everyday lives. Business leaders must get accustomed with this technology now. There is, as with any new technology, a risk that will be applied to something that simply doesn’t need it, and money will be wasted, as Linthicum argues. Education is key to this, and the understanding that just because vendors are trying to push a product, doesn’t necessarily mean they’re not telling the truth.