Machine learning adoption thwarted by lack of human skills

We’ve all heard the dire predictions about robots coming to steal our jobs. As technologies such as machine learning, AI and automation advance by the day, workplaces everywhere are being transformed; naturally, some people fear that they will become redundant — depending on their job, some may be right.  In this age of AI and ML, ambivalence is ripe; but something ironic has emerged: when it comes to advancing these cutting-edge technologies, the lack of human skills and knowledge is slowing innovation down.

In a new survey by Cloudera, the software firm, exploring the benefits and roadblocks of ML adoption across Europe, 51% of business leaders said that the skills shortage was holding them back from implementation. According to Cloudera, companies are eager to use ML — it’s second only to analytics as the key investment priority for businesses; ahead of other disciplines like IoT, artificial intelligence and data science.

“Although most IT buyers understand the benefits of machine learning, with 33% of respondents saying they have already seen tangible ROI from its use, many are still unsure about how to implement and how it will impact their businesses,” said Stephen Line, VP EMEA at Cloudera. “In what is still the early stages for many businesses in actually implementing ML, it’s unsurprising to learn that the skills gap and investment are key factors in preventing many companies from using it to improve efficiency and drive growth. That said, with the benefits of ML quite clear, the race is now on for businesses to overcome their barriers to deliver a better experience for their customers.”

Here’s a solution to the AI talent shortage: Recruit philosophy students

A partial solution to the problem of lack of artificial intelligence talent might lie with recruiting philosophy students. Read here

A new profession

Similar to data science, ML is progressing in a distinctly different way from other job markets. Because ML rotates around gathering, collating and interpreting data, it traverses numerous disciplines; maths, statistics, and programming are all required. It’s difficult to write this in a job description let alone actually find it.

As you can imagine, ML is pretty complicated stuff, and it’s not something just any old computer engineer can grasp. ML requires cream of the crop computer scientists who can deal with large volumes of data at scale.

A natural intuition for maths is essential. This contrasts, with traditional software developers, who  don’t need to be that great at maths thanks to the availability of maths libraries and other functions that relieve them from doing equations the hard way. With ML, a developer needs to grasp complicated maths such as linear algebra, calculus and gradient descent.

As we all know, today there’s a dearth of skills in all areas of STEM. Information Age recently reported how 94% of business leaders surveyed by OpsRamp are having a  “somewhat difficult” time trying to find candidates with the right technology and business skills to meet digital transformation goals.

Competition for this shallow pool of candidates is fierce, and the arrival of new roles is outstripping supply.

ML, has created a new profession. Interestingly, people are finding their way into it through unconventional routes. According to a study from data scientist community Kaggle, the vast majority of employed machine learning specialists today gained their skills by way of self-learning (27%) or a Massive Open Online Course (MOOC) (32%). Only 20% of people got their start in data science at a university, while 18% majored in math/statistics.

How automated machine learning is helping to bridge the AI skills gap

Dr Ken Sanford, the US lead Analytics Architect for Dataiku, discusses how AutoML will help solve the AI skills gap

A creative approach is needed

This approach to self-learning has come with its disadvantages; the diverse range of courses has led to confusion. Given how broad the ML domain is and how fast it changes, no course is really comprehensive enough, leaving developers to deal with overlapping and conflicting lessons. It’s also challenging for recruiters who struggle to find truly certified talent.

Many companies that have successfully built an ML department has done so through setting up specialised training programs. This way a company can attract people with the core abilities needed and steer the development of their skills in the direction of the business.

For more senior roles, qualified ML developers are often found through network connections, academic papers and conferences.

Although these recruitment practices have been seen before in tech, they do depart from more traditional means such as university partnerships and outsourcing.

Compared to other professionals, the way people are becoming ML professionals is different. Likewise, business leaders will need to attract and recruit them in a different way; this will involve creativity.

Put your money where your mouth is: How to make the most out of your machine learning investments

How to make the most of machine learning investments, including 8 immediate actions to help remove barriers and move the needle with ML strategies

Avatar photo

Andrew Ross

As a reporter with Information Age, Andrew Ross writes articles for technology leaders; helping them manage business critical issues both for today and in the future