Microsoft Research director fears 'potential misuse' of AI technologies

Eric Horvitz admits artificial intelligence technologies pose threats.

Artificial intelligence and machine learning have been at the core of Microsoft Research in the last few years as the company believes that the future of the firm and the humanity, in general, relies on the technology's advancements. With the great power that comes with AI technologies, it poses a great amount of threat to the humanity as a whole.

BBC Newsnight's technology editor David Grossman visited Microsoft's Building 99, the company's research facility in Redmond, Washington that houses its artificial intelligence unit. In the episode, Microsoft Research managing director Eric Horvitz asserted: "We're betting the future of the company on advances in AI. We think this is the future."

Horvitz, however, expressed his qualms over the imminent risks of this technology once in full swing. "I'm concerned about the potential misuse of this technology by malevolent forces, by people with ill will, by state and non-state actors who can gain strong powers with these technologies," says Horvitz.

Despite the imminent threats that come along with artificial intelligence, Horvitz was quick to add that artificial intelligence is also the solution to this problem. Some AI software will act out as the defence against perpetrators.

Horvitz was straightforward to admit artificial intelligence technologies can also play a crucial role during elections and democratic processes, adding that authoritarian regimes can largely benefit from it for tracking and surveillance.

"AI systems can be designed to persuade in algorithmic way, to optimise goals of changing someone's beliefs or enhancing the beliefs of one thing versus another," says Horvitz. More importantly, the scientist sees AI as an open ground for more discussions and collaborations to turn all these risks into a positive influence.

In the meantime, the episode tackled Microsoft's Bot Framework as the future will see humans talking a lot to computers. XiaoIce, the company's social media chatterbot in China, interacts with humans to gather more information of their individuality so it understands the humanity more. Microsoft's principal architect Dan Driscoll said humans have been smitten by bots. In fact, Chinese users engage to XiaoIce up to 23 turns per conversation so they "form emotional relationships" with the chat bot.