SELF-LEARNING SYSTEMS THAT MAKE DECISIONS. COMPUTERS THAT INTERACT WITH PEOPLE. THANKS TO ARTIFICIAL INTELLIGENCE, BUSINESSES ARE IMPROVING THEIR SERVICES AND INVENTING COMPLETELY NEW MODELS. THE TECHNOLOGY IS HERE, BUT THERE ARE STILL MANY MORE OPPORTUNITIES TO BE DISCOVERED.
There’s not much consensus about the definition of artificial intelligence. There are a great many links to other areas within IT – to the Internet of Things (IoT), for data collection, but also to the use of algorithms for data analysis, and to machine learning.
When we talk about the use of artificial intelligence in the business world, we are usually referring to self-learning systems that employ data analysis and previous decisions to make predictions and support decisions. In this way artificial intelligence allows a business to evolve from the pure analysis of historical data to insights in real time and to predictive analysis.
“We prefer the term advanced intelligence,” says Cédric Mulier, Cognitive Solutions Leader Benelux at IBM. “It refers to an application with added value, applied by a learning and interacting solution that helps people to make better decisions.” Whatever the accepted definition, most large businesses are thinking about the phenomenon. Those with large amounts of data know they are sitting on a potential goldmine.
“Many businesses have that data,” says Stéphane Jacobs, Director Mobility Payment Solutions at Be-Mobile, “but they aren’t always able to match their goals to the right applications.” Be-Mobile is already demonstrating how it can be done. Stéphane Jacobs: “We combine data on traffic from 300 different sources, analyze it and sell the results to, for example, car manufacturers, who use it to intelligently navigate drivers through traffic.”
“Historical data of companies can be an obstacle that gives new players the chance to enter the market.”
– Cédric Mulier, Cognitive Solutions Leader Benelux at IBM
In the world of artificial intelligence, applications are obviously not limited to providing traffic information. Stéphane Jacobs: “The next step is advice. For example, the application can then advise you to work from home for a while, because of the congestion, or take the train so you get to your appointment on time.” This is where the self-learning aspect of artificial intelligence comes into its own. The suggestions are the result of the analysis of historical and real-time data – and of the results that these provided in the past.
Cédric Mulier: “That’s also what we see with the IBM application Watson Health at the UCB. The computer processes and analyzes medical images and data and in 70% of cases can give the right diagnosis – and recommend treatment. With each new diagnosis the machine learns a little bit more, so that it can continue to suggest better diagnoses and treatments.”
Applying artificial intelligence in the business world means looking for use cases. But where are they located? Should companies focus their efforts on concrete results, such as cost reduction or better customer service? Or will artificial intelligence deliver completely new business models?
Cédric Mulier: “Both are possible. For example, the elevato manufacturer KONE uses an IoT solution with sensors that collect data in order to monitor how well the lifts are working and to predict when maintenance is required. At the same time the company gets information about how many people use the lift, at what times, etc.”
This is the point at which a rationale for new services emerges. “That’s right,” says Philippe Van Impe, founder of the Brussels Data Science Community, “but that will only work if the company already has some understanding about what is possible in the area. And most businesses are not yet in that position. First of all, there is a need for more awareness about the possibilities of artificial intelligence.”
Once there is insight, the opportunities are there for the taking. Stéphane Jacobs: “Traffic information is not just interesting for people sitting in traffic. Our information is also useful for businesses who advertise on the billboards along busy main roads. The density of the congestion is something that helps determine the billboard price.” This example shows once again that it is not the data that deliver the added value, but the analysis of them.
“We also see a clear difference between greenfield and legacy,” says Jean-Marie Stas, Marketing Manager at Proximus. “Banks, retailers and telecoms companies have a lot of data about their clients, but they can’t just ignore the existing processes and structures.”
Cédric Mulier: “That history can be an obstacle, creating a relative inertia that gives new players the chance to enter the market. As a result, you get telecom companies and car manufacturers that start offering banking products, or digital businesses that have a strong influence on the market shares and margins of retailers or hotel businesses. That said, a great many historical players do introduce the necessary organization and culture to innovate with new models and new added-value services that benefit from the (internally and externally) available data and learning systems.
For that reason, it’s crucial to determine the so-called ‘incubators,’ data officers or targeted innovators who deliver reports to the board of directors.” The crucial ingredients for continuing to build on their historical advantages are their brand name, the know-how of their employees and their credibility among their contacts. Cédric Mulier: “The North Face communicates with its customers via an intelligent online solution that helps them choose a jacket, depending on their preferences and the context. That choice is then brought to a conclusion with the sales person.”
“By combining data sources – from your diary, your location, the traffic – artificial intelligence will soon be providing you with a solution for seamless travel.”
– Stéphane Jacobs, Director Mobility Payment Solutions at Be-Mobile
In this context the arrival of the General Data Protection Regulation (GDPR) also plays a role. The new legislation requires businesses to take responsibility for data management. “The crucial thing is that company management is now getting involved in this,” says Philippe Van Impe. “New insight is needed about what is allowed and what is possible, both legally and practically. At the same time, it remains a challenge. Data and privacy are essential but not necessarily sexy topics.”
And here too, the contrast between greenfield and legacy plays a role. Philippe Van Impe: “Startups don’t have any historical data; they can start with a clean slate. For existing companies – with a history of data and processes – the situation is completely different.”
The technology in itself should not be an obstacle and there are solutions based on various data types. Cédric Mulier: “The biggest challenge lies in the development of a correct vision without conceptual errors: ‘think big, start small’ and be pragmatic so you can adapt quickly (‘agile’). You have to find the right use case, one that can create a certain dynamic in order to generate a snowball effect.” In practice it comes down to releasing the necessary resources for it.
Philippe Van Impe: “My advice is to start simple, with a limited dataset and a simple algorithm. That should already provide more artificial intelligence than most companies currently possess.” Once you’ve started, you can gradually expand the use of intelligent solutions.
“Where does that lead?” wonders Jean-Marie Stas. “Will our boss soon be a computer? Will the system soon be telling us what to do?” In part yes, although the computer will not have the last word. Cédric Mulier: “The machine gives advice with a certain probability, but the human will take the final decision. The computer can give a diagnosis, but the doctor decides.
” We will probably see a lot of direct added value initially in banking, insurance and retail, where the computer can give a faster and more correct offer for an insurance product or can answer a frequently asked question. “In the end it’s about the customer experience and the service offered,” concludes Stéphane Jacobs.
“By combining data sources – from your diary, your location, the traffic – artificial intelligence will soon be providing you with a solution for seamless travel. Then the application will also give you real instructions: leave now if you want to get to your meeting on time. And there will be a self-driving car waiting to take you to your destination. As a first step towards the self-driving car, today’s vehicles can talk to each other and to the road infrastructure.”