Artificial intelligence, in addition to a multitude of useful things, can pose a great danger, from automatic hacking, through attacks on individuals to manipulating elections, experts said in the report “Abuse of artificial intelligence”.
Sharar Avin of the Cambridge University Center for Research on Exterior Risks told the BBC that the report focused on artificial intelligence options that exist now or will be available within five years, not in the distant future.
The report’s authors warn that governments must consider new laws, and that those working on artificial intelligence systems must work more on preventing technology abuse.
Avin said that in the near future, technologies such as Google AlphaGo, which was the first to win a man in the game, could use hackers to find patterns in the data and codes.
Among the dangers is that individuals can buy drones with face detection software and thus attack their targets.
Avin said developers could automatically make “fake” videos of political manipulation, while hackers could use speech synthesis programs to attack political opponents.
The report states that bots are already being used to manipulate social networks that control people, which has led to the impact of political reporting on mainstream media and the formation of political attitudes.
Sophisticated artificial intelligence systems could target groups with the right message at the most appropriate time in order to maximize the potential for convincing.
In addition to manipulating elections, it can also be used to recruit terrorists, while authoritarian regimes will receive additional control mechanisms that are unlikely to be available in democratic systems.
Michael Brandey of the Institute for the Future of Humanity at Oxford University said that artificial intelligence will change the “landscape of hazards for citizens, organizations and states.”
“It is often the case that artificial intelligence systems not only reach the level of human results but significantly outperform them,” he said.
Brandeidge added that it is necessary to consider the implications of superhuman hacking, surveillance, persuasion and identification of physical targets.
The report “Abuse of artificial intelligence”, which has hundreds of pages, identifies three areas – digital, physical and political – in which artificial intelligence will most likely be abused.
About Google TranslateCommunityMobileAbout GooglePriv