Ethical, unbiased artificial intelligence against fraud in utility supplies and services

Bringing abuse and fraud under control in services like electricity and gas is essential to optimize resources, detect irregularities and adopt measures to ensure that systems run smoothly. This is the purpose and objective of Fraud Research, a tool based on big data, with ethics into account, which detects anomalies and abuses in this type of services. This project has been finalist in the ninth edition of SpinUOC, the the Universitat Oberta de Catalunya (UOC) event to foster entrepreneurship. Fraud Research is a spin-off of the Universitat Politècnica de Catalunya (UPC).

"We live in a digitized world, so we leave a digital trail. This includes interaction between companies and their customers. Accordingly, these data can be analysed by the company with artificial intelligence (AI) to understand these customers better and, therefore, detect whether they are committing fraud or there is another reason for their anomalous behaviour," said Bernat Coma, one of the researchers behind this project and a course instructor in the Faculty of Computer Science, Multimedia and Telecommunications at the UOC and associated professor and researcher of the Process and Data Science group at the UPC – BarcelonaTech.

Specifically, Fraud Research provides companies with the knowledge developed by a group of experts to detect fraud and other anomalies thanks to the application of artificial intelligence and other data science techniques. "Our solutions are based on two concepts: a high level of precision and the ethical use of artificial intelligence," stated Coma. 

Ethical artificial intelligence 

In this sense, it is important to remember that in recent years the use of artificial intelligence has led to a number of media scandals and problems for organizations and companies because the algorithms reproduced certain behaviour patterns or biases that were socially unacceptable, such as racism and sexism. This is due to the existence of black box algorithms, which are processes that are in principle the best at making predictions with large amounts of data, but they are not interpretable, that is, the patterns they use to obtain data are unknown. 

"Our aim is to ensure that our algorithms do not reproduce such biases in their predictions, given that if AI is biased, it is not intelligent," explained the specialist, who has developed these algorithms along with a group of experts over the course of several years with the aim of achieving artificial intelligence that is fair, precise and ethical without suffering a loss of efficacy or effectiveness.

The vast majority of information and data in today's society are computerized. This is why solutions for today's and tomorrow's problems should be based on big data analysis. "The technological evolution that allows us to study enormous quantities of data in a reasonable period of time makes it possible to consider new technological solutions to automate processes," said the expert. 

In Spain, it is estimated that electricity fraud amounts to more than 4 billion kilowatt-hours per year, a volume of energy that is capable of supplying a city of more than one million inhabitants. This crime generates estimated losses of more than €150 million per year, resulting in higher electricity bills for all consumers.

According to a study conducted by Northeast Group, fraud in this type of services on the global scale caused by stealing public services, tampering with meters, bypassing meters, defective or broken meters, supplies that are not measured and technical and human errors in meter readings amounts to $96 billion a year.

"More than 49% of organizations currently operating have suffered fraud in recent years. At Fraud Research we believe that by using artificial intelligence we can reduce these losses. This is why we have presented our fraud management software tool," stated the researchers behind this project. 

Applications in services 

Although for the moment, Fraud Research is applied to the detection of the aforementioned cases of abuse and fraud, the algorithms that have been designed and the technology that has been developed can be applied to any field that includes data and requires analysis to meet the needs of a specific process. 

"We tend to cite fraud and anomaly detection in electricity, gas and water companies to explain the tool, but it could also be very useful for banks to determine whether a potential customer will pay back a loan and even for public institutions and universities," said Coma, who stated that artificial intelligence could also determine whether a student would pay for their studies or whether a citizen is putting a public subsidy to good use. 

This project, which was a finalist in SpinUOC, a programme fostered by the Hubbik platform, promotes Sustainable Development Goals (SDGs) 8, Decent Work and Economic Growth, 9, Industries, Innovation and Infrastructure and 12, Responsible Consumption and Production.

UOC R&I

The UOC's research and innovation (R&I) is helping overcome pressing challenges faced by global societies in the 21st century, by studying interactions between technology and human & social sciences with a specific focus on the network society, e-learning and e-health. Over 500 researchers and 51 research groups work among the University's seven faculties and two research centres: the Internet Interdisciplinary Institute (IN3) and the eHealth Center (eHC).

The United Nations' 2030 Agenda for Sustainable Development and open knowledge serve as strategic pillars for the UOC's teaching, research and innovation. More information: research.uoc.edu. #UOC25years