Some fear the age of Robots, AI and Deep Learning. But with adequate governance, policies and strategic planning, these fears are largely unwarranted. Organizations need to understand and mitigate the risks associated with AI, robotics and cognitive technologies rather than avoid them.
Bots gone wild: How to manage Artificial Intelligence (AI) risk
Back in July, news headlines across the globe announced every sci-fi fan’s worst fear: two Facebook bots were caught talking to one another in their own language. A language humans couldn’t understand.
It all started when Facebook researchers attempted to teach chatbots the art of negotiation through a simple game which involved dividing a collection of objects between themselves. At some point during the negotiation game, two bots stopped using recognizable sentences—they still used English words, just not in the right context. Despite the weird language, the bots were able to continue negotiations—and score points in the game.[i]
Deloitte Senior Manager, Umang Handa, has built a career in this space, with a particular focus on cybersecurity, analytics, and robotic process automation (RPA). Deloitte’s Cyber Risk services practice is the largest integrated cyber risk management practice in the world, helping leading organizations build secure, vigilant, and resilient risk strategies.
Handa notes that, given the rapid evolution of AI in recent years—and the continuous warnings expressed by the likes of Bill Gates, Stephen Hawking, and Elon Musk, among others[ii]—it’s understandable the media would pick up on such a story. And while those close to the situation believe what happened was more of a failed experiment than a sign that robots are conspiring to destroy the human race, it does highlight the risks associated with AI.
Imagine, for example, these bots were acting as bank representatives when they strayed off-course. Perhaps, rather than speaking a different language, they moved money to the wrong account—or made a decision to issue more money than requested. The point is, the Facebook story points to a larger issue—while AI and machine learning come with a plethora of business benefits, a computer’s ability to gain additional capabilities and powers can also, occasionally, run off-course.
The World Economic Forum agrees. Despite the unquestionable opportunities presented by intelligent automation, these emerging technologies also present potential threats (see figure 1).
That’s why it’s important that businesses employing these types of tools mitigate the associated risks. According to Handa, this means stepping up governance surrounding the bots’ design to ensure they’re properly architected and deployed; implementing appropriate policies to ensure the bots’ activities are centrally controlled and monitored consistently; and actively managing your digital workforce.
While AI is evolving rapidly, it isn’t anywhere close to going rogue like HAL 9000 from 2001: A Space Odyssey. That said, to get the most out of your digital investments—and ensure they move your business forward, rather than cause damaging side-effects—it’s integral to have an end-to-end security solution in place.