AI2 Was Created by Humans and It Evolves! Will It Rebel?


According to a recent report by the World Economic Forum, more than five million jobs could be lost by 2020, as a result of advancements in artificial intelligence (AI), robotics and other technological changes. Now, even the task of detecting cybersecurity threats could be assigned to an AI, albeit initially with the help of human analysts.

Scientists at Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Lab (CSAIL) and startup company PatternEx developed a system called Artificial Intelligence Squared (AI2) that uses “active learning” to detect 85 percent of cyberattacks—approximately three times better than previous benchmarks. AI2 achieves this impressive feat by first combing through data and flagging suspicious activity. Then, the results are presented to a human analysts, who confirms actual attacks. Then, AI2 incorporates the feedback into a “supervised model” to be used to analyze the next set of data and this process is repeated.

CSAIL and PatternEx tested AI2 on 3.6 billion pieces of data generated by millions of users over three months. According to PatternEx Chief Data Scientist Ignacio Arnaldo, AI2 “continuously generates new models that it can refine in as little as few hours, meaning it can improve its detection rates significantly and rapidly.” 

As AI2 evolves, it will become better at detecting cybersecurity threats. Hopefully, it won’t evolve into conscious, sentient “beings”—such as Skynet, HAL 9000 or The Cylons. Or else, by 2020, the loss of millions of jobs may be the least of humanity’s concerns—AI2 was created by man. It evolved. It rebelled. There are many copies. And it has a plan.

To keep up with the constantly evolving world of privacy and security sign up for the Bloomberg BNA Privacy and Security Update.