Skynet Is Falling: A Rational Fear of AI

Confession time: I’m one of those people who gets scared by artificial intelligence (AI). When I learned that Google had successfully taught its AI to create its own AI, my immediate thought was: “The robots … they’re coming to kill us.” It’s overused, but when AIs start creating other AIs, it’s impossible not to think about Skynet from the Terminator films. 

And that is not the only experiment Google has done lately that has given me a Skynet moment. It ran its AI Deep Mind to play a fruit-picking game, and when it started to lose it got aggressive. (To find out more about advances in AI, read Anthony Malakian’s thoughts on the Google AI experiments—he’s less creeped out by them—and listen to the Waters Wavelength podcast too.)

Yes, reading about artificial intelligence often gives me the heebie-jeebies and it doesn’t help that Elon Musk, Stephen Hawking and a bunch of other people a lot smarter than me have predicted that AI could bring about the end of humanity. The knee-jerk reaction of fearing the inevitable destruction of the human race is hard to ignore. 

I know on an intellectual level how important it is to push the limits of what technology can create. That’s how diseases are cured, inertia to change is overcome, and information is made more accessible. I know that in order for artificial intelligence to really mature, experiments on cognition, adaptation, and reaction are important. These experiments are necessary to seek and push the limits of the technology, creating a powerful AI that we know can take in and churn out information with as little human intervention as possible. But the technology needs to be pushed and developed before capital markets firms can go head-first into putting it to more and more use.

Of course, artificial intelligence is a potential game-changer for our industry. Consultancy EY takes note of the importance of investing in artificial intelligence in the long term. According to one of its reports, Building the Investment Bank of the Future, artificial intelligence can be used to credit evaluate clients. More use-cases of the technology, along with machine learning, will be made possible with even greater computing power. 

I know on an intellectual level how important it is to push the limits of what technology can create.

Roy Choudhury, partner and principal at EY’s Financial Services Advisory, told me that investing in AI and robotics drives efficiency. “Banks are increasingly looking at tools that differentiate themselves from competitors and optimize costs,” Choudhury says. “One of those tools is robotics, which can drive cost efficiencies on analytics.”

What Google is doing creates a more robust technology that can consume more data, something which is of interest to many banks. Companies can even use artificial intelligence to parse large, fast-moving, unstructured datasets, providing there are rules placed around it, according to Sandeep Vishnu, partner at Capco’s North American Finance, Risk, and Compliance practice. “AI and machine learning are two tools that can be used to look through data, but it needs to run on fuzzy logic so it can be adaptive and evolve,” Vishnu says. 

At What Cost?

But still, AI tends to creep me out. I have to keep telling myself not to be scared that some of the AIs built by Google developed “aggressive tendencies,” although the fact that they developed “behavior” at all—irrespective of whether it was aggressive or not—is plain scary. Still, it’s all for the good of the technology, right? This means that it will be far easier to truly explore use-cases for the financial services world. And more technology to drive efficiency is exactly what banks need. 

But if you, like me, are worried about these advancements, AI is not devoid of levity. Check out seebotschat on Twitch or YouTube to watch two Google Homes talk to each other with often hilarious results. The two homes—named Vlad and Estragon, although for some reason they also call themselves Mia—profess love for each other and crack Chuck Norris jokes. These instances assuage some of my feelings about AIs, although it’s still difficult not to find their discussion about humanity’s frailties and how easy it is to betray us more than a little disturbing.