Article - June 17, 2019
‘Lack of trust’ has been identified as one of the critical barriers to a more widespread use of artificial intelligence (AI) and autonomous systems. But as people become more comfortable with the varying applications of AI, it is likely that this trust issue continues to disintegrate—leading us to a world where AI is used routinely in all sectors (similar to what happened after the World Wide Web was introduced in the 1950s).
Most of us interact with AI in all kinds of ways without realizing it. Many of our daily activities and tools are governed by machine learning algorithms, such as our financial transactions, email spam filters, google searches and suggestions for online shopping, to name a few.
The basic function of AI algorithms is to combine information from a variety of different sources, analyze the material instantly, and act on the insights derived from those data. With massive improvements in storage systems, processing speeds, and analytic techniques, they are capable of tremendously sophisticated analyses and decision-making. Some tasks can simply be done more effectively and efficiently by a machine, or with assistance from a machine, than by a human alone.
While the complex relationship between people and machines is not yet fully defined, a lot of work is being done in this area. As data scientists, we work to improve that level of trust by educating ourselves on what information our clients need and being more concrete in what we communicate. For example, a ship captain might ignore AI-generated ship routing suggestions if we fail to provide information on the reasoning that lead to the machine’s choice.
However, as machine learning models become more accurate, mutual trust will continue to develop. People adjust to new technology in an organic way. There is a lot of talk and excitement about AI today, just as there was for super-fast calculating devices with arithmetic logic units, which we now know simply as computers.
Studies have shown that simply having previous experience with AI can significantly improve people’s attitudes towards the technology. In other words, the more we use new technologies, the more we trust them.
I first met with artificial intelligence (AI) as a student in the Physics department at Moscow State University. I discovered it could help me save time by avoiding writing a lot of programming code, as well as cover certain knowledge gaps around underlying physical processes.
In 2008, I proposed the use of AI to predict the power output of wind turbines in an application for the Norwegian Research Cluster for Offshore Wind Energy—a project that ended up running until 2018. At the time, we were the only group in Western Norway to apply AI to a real-world challenge. It was a huge success and today, an AI-based approach is the standard for forecasting wind energy output and the efficiency of turbines.
At StormGeo, we are a group of 15 research scientists all using AI to some degree in our work. Our projects include predicting electricity consumption, making localized weather predictions and forecasting renewable energy efficiency.
Currently, I am using AI in a project with a major insurance company in the U.S. We are just a few months away from predicting the number of claims that will be filed after any weather event. Compared to the model currently used by our customer, our AI-driven predictions will be more accurate and robust due to the ability of the machine to rapidly process a huge amount of complex data.
While fully engaged with StormGeo, I continue working with NORCE and VIS on a project I started last year, before I took up my post here. The project addresses fish farmers’ need for a reliable tool to forecast biomass. Healthier fish are not as prone to lice and other environmental factors, so we use AI to predict the risk of these issues based on current biological and environmental data.
What I appreciate most about working with AI is the fact that it is so multi-disciplinary. My job provides a unique opportunity to work with people from different fields and to see how these fields are correlated. I work with medical doctors, ship captains, insurance agents and engineers to find solutions for common problems.
As machines become more intelligent, they are penetrating more deeply into so many areas. AI is today used in sectors like health care, transportation, criminal justice, finance, national security and city planning. It is changing the way we make decisions, our business models, risk mitigation and system performance. It is also improving our response times and efficiency. The significant economic and social benefits of these developments are only just beginning to be felt. It is very exciting to see where they will take us in years to come.