The world is agog with the looming reality that artificial intelligence (AI) will become part of everyday life, but, according to a new study from Belatrix Software, although plenty of companies are interested in adopting some form of AI or machine learning solution, in reality very few have taken the plunge.
Conducted in November 2016, the study — entitled Powering the Adoption of Machine Learning – found that there was huge curiosity about the technologies driving AI and machine learning, but also that companies are struggling to understand how to get started with it. Questions in the survey were open-ended, resulting in a variety of answers from the 72 companies who participated.
“The idea behind the study was to ask people about their interest in machine learning, whether or not they think it’ll have a big impact on their organization, whether or not they’ve already started an initiative, and what have been some of their challenges,” said Charles Green, Director of Thought Leadership at Belatrix Software, who authored the report.
The study found that 81% of respondents believe that machine learning will have some impact or a significant impact on their organization in the next five years, with the biggest repercussion being improved operational efficiency (53% of respondents answered with this).
Other expected impacts are the automation of business processes (50%); faster responses to market changes and customer desires (46%); improved sharing of insights and information throughout the organization (43%); and lower security risks and improved responses to security threats (28%). Only 10% (or 1 person, according to Green) of respondents believe that it will have no impact whatsoever, while 10 people said it would have little impact.
Getting Started with Machine Learning
When asking if companies had already started a machine learning initiative in their organizations, the answers revealed that only 18% had done so, 40% were investigating it but hadn’t started, and 43% had no plans to start one at all.
“The gap between what companies want to do and what they are doing is very interesting,” said Green. “It’s clear that the interest is there, but taking the next step and getting started seems to be the main issue.”
The inability to take this first foray into machine learning comes down to a number of factors, including the belief that the technology is not mature enough, difficulty in developing the right business case for investment, difficulty managing and analyzing the data, and simply not understanding why the technology is important. But the biggest challenge is the difficulty in finding people with the right skills, something that is echoed across the region in the IT sector in general.
“It’s a huge challenge to find data scientists, people with machine learning experience, or people with the skills to analyze and use the data, as well as those who can create the algorithms required for machine learning,” said Green. “Secondly, while the technology is still emerging, there are many ongoing developments. It’s clear that AI is a long way from how we might imagine it. Even so, there are very clear case studies that demonstrate where machine learning can be used to provide a lot of value, and it’s maturing rapidly.”
Luis Humberto Rojas, Business Developer at Nearshore Delivery Solutions, concurs: “I believe that few corporations are aware about AI and its use in day-to-day operations. The information out there is not enough for them and is difficult to understand. Companies still think “how can I implement something that could be from a sci-fi movie into my business?” To change this, there needs to be more companies that are passionate about AI and can get the right information out there, while implementing AI solutions in different industries.”
Current Initiatives in Place
Machine learning can be used for a vast range of different issues, which was highlighted in the study when asking what current initiatives companies were working on. One respondent said that they were using it to improve collaboration between teams, another is using IBM Watson for technical support, one is using implementing predictive models to better serve customers, and another wants to use machine learning to improve search and responsiveness.
“While some companies are using services like IBM Watson, there are others creating their own machine learning algorithms in order to find their own particular niches that they can provide customer value,” said Green. “One of the areas that is driving the adoption and maturity of machine learning is big data. There is so much more data that can be analyzed and fed into machine learning algorithms to train these systems, but it is also much cheaper for companies to store data than it was a few years ago. Open-source solutions are certainly a key factor in this progression.”
“Sure, AI and machine learning is not easy to implement, but there a plenty of AI tools available,” said Rojas. “Companies may need specialized talent, but the tools are already out there. Major tech companies are already developing their AI practices — machine learning with Google and Tensorflow, industrial internet with GE Predix, and cognitive computing with IBM Watson, to name a few examples.”
Working Examples of AI
Belatrix is itself looking at machine learning and artificial intelligence as a strategic priority for 2017. The firm has established a team — led by Alex Robbio, President and Co-Founder – to research the technologies and create proof of concepts, while also working with clients on the topic.
One example of their progress is a proof of concept for a real-estate company. The technology collects thousands of data points from the company’s Internet-of-things (IoT) devices installed in its buildings every second, resulting in huge quantities of data. “The objective here is to use machine learning to help them operate their buildings more efficiently,” said Green. “One example is how they can better use and predict their energy use, or reduce maintenance costs. This brings together IoT, machine learning, and the different types of data that come from a variety of sources.”
Belatrix also has an internal project that uses neural networks to predict the risk of personnel attrition. The system can work out if people are likely to leave the company, using just the information that already exists in HR, project management, and training systems.
Before adopting AI or machine learning, Green has one piece of advice to follow: “Start small. You can’t jump into a full-blown machine learning implementation from nothing, so look to hire data scientists, or work with partners who have access to the right skills. In many cases, this can start off by exploring the data, identifying what you can do with it, and finding a niche or project you can start creating that will delight your customers – it becomes much easier to show the value of that to your clients.”
Similarly, Rojas believes that companies should tread carefully at first: “My advice is to get to know AI right now and stop looking at it as something from the future. Second, understand the tools that are already available and determine how they can be applied to your existing business structure. Third, reach out to companies that are dedicated to the technology and look for their advice on how to build the right synergies.”