Japan develops a system for predicting crime using AI and large data
Back in January this year, the authorities of Kanagawa Prefecture announced plans to be the first in the country to introduce a system for predicting crimes. In order to already have an experimental network of data collection and processing in the whole region for the Tokyo Olympics 2020. Now the concept has an ideological leader, an analyst Mami Kajita, and the police are discussing with the authorities details of the lease of supercomputers and the involvement of several AI.
Mami Kajita, a mathematician physicist and weather forecasting specialist, while on probation in Italy, was faced with the fact that she was constantly robbed. The foreigner decided to apply her skills and developed an algorithm for analyzing police reports, taking into account such data as terrain, time, weather and other factors. As a result, she received a program for predictions and changed her travel routes around the city, ceasing to be a victim of criminals.
Khajita argues that in the United States such prognostic systems have been used secretly by the police for a long time. She also wants to create an application for civilians that will help them minimize the risk of getting into trouble. But this requires extensive initial data, and in the format of “big data”, plus the processing power for processing them. All this is in the hands of the Japanese authorities, and the police are willing to provide information from cameras, sensors, reports and directly from the streets, from patrols. Questions – in the technical implementation of the system.
For example, AI can easily read tens of billions of tweets and predict the meeting of twenty thousand people in the park for public reading of the poems. There will be rainy weather, twilight, many foreigners – a paradise for pickpockets. Here he is, the prediction of the crime, but how exactly should he react to the police, and how to take into account the influence of the reaction on the further development of events, so as not to get worse? And if at the same time a similar forecast comes to smartphones of thousands of people, then will the situation become completely confusing? In a word, there is still a lot to be done.