Machine LearningMachine Learning (ML) may have been claimed as the next best thing, but it still is yet to be implemented in mainstream enterprise adoption.

Predictors claimed that 50% of organisations had joined the queue to transform themselves this year with Machine Learning, those more sceptical about it put the number closer to 15%.

Despite ML and AI has the potential to reshape enterprise computing, a lot of hurdles still remain. The reality is that big data is difficult. While hope remains high, analyst Nick Heudecker claimed, “Only 15% of organisations get to productions”, and this will “likely be much lower with ML.”

However, hope remains high with a response to a Belatrix Software survey stating that 81% of respondents proclaiming that “ML will have some impact of a significant impact on their organisation in the next five years.” But although conveying a positive outlook, only 18% of those companies had bothered to get started, while 40% are kicking the proverbial tires, leaving 42% who have done nothing.

A lot of the lack in action most probably comes down to the massive gap between ML myth and reality. David Beyer from Amplify Partners, explained, “Too many businesses now are pitching AI almost as though its batteries included.” This is dangerous because it leads companies to either over-invest (and then face a tremendous trough of disillusionment), or to steer clear when the slightest bit of real research reveals that ML is very hard and not something the average Python engineer is going to spin up in her spare time.

One of the gating factors to ML success is data. To properly train models, an enterprise needs “an unearthly amount of data” as Neil Lawrence, a member of Amazon’s AI team and professor of machine learning at the University of Sheffield, puts it. More than any mind-blowingly great algorithm, he says, “progress is driven far more by the availability of data than an improvement in algorithms”.

Few enterprises have such large amounts of data; even if they get past each of the challenges they will often fall short when it comes to people. Most companies are happy to talk about being data driven, but precious few actually are.

It has been suggested that ML would be more manageable if people knew how to model it, but there is a real lack of experts. Gartner analyst Merv Adrian believes the biggest reason for ML’s poor success rates is missing skills. According to Ben Lorica and Mike Loukides, it is so difficult to find more data driven people because ML is a “practical discipline, not really something easily picked up in the classroom setting.”

To conclude, it would seem that most of what get discussed as ML usually isn’t. Basecamp data scientist Noah Lorang, commented, “There is a very small subset of business problems that are best solved by machine learning; most of them just need good data and an understanding of what it means that is best gained using simple methods.”