According to the New Scientist publication, very soon artificial intelligence will watch for violators who steal electricity. The AI algorithm was developed by experts from the University of Luxembourg and has already been successfully tested in Brazil.
Brazil was not chosen for experiments by accident. The fact is that in this country the issue of electricity theft is especially acute: according to official statistics, up to 40% of consumers twist meter readings. But more developed countries are not immune from this phenomenon either. For example, in the UK, electricity meter fraud results in a loss of £ 440 million annually.
A group of researchers from Luxembourg have tested their algorithm on 3.6 million Brazilian homes for 5 years. Every month, scientists collected data from meter readings, on the basis of which an algorithm of actions was developed, which was “fed” to artificial intelligence. This algorithm is able to recognize when the energy consumption in the home was suspiciously low, but the electricity was being used. It turned out that AI correctly detects almost all potential cases of fraud or erroneous measurements, which may indicate breakdowns of meters and even violations in the power grids themselves. As stated by Paul Ruisvelt of the University of London Energy Institute,
“Interestingly, this AI can accurately determine the fact of theft even by people who steal only 10% of the electricity. However, this algorithm is not always applicable. For example, it is normal for adjacent buildings to have very different levels of energy use, which can lead to false positives.”