13 Jul 2018 AI is not sexy at all

Forget all the miraculous things you’ve read about artificial intelligence in magazines. Machines defeat the world’s best Go players. Machines beat poker professionals, write poems, and outperform humans on IQ tests. Machines might even one day eradicate cancer. Mind-blowing, isn’t it? But so far from reality.

Believe it or not, AI is incredibly boring. In fact, AI infiltrates our daily lives without us even noticing.

For example, AI is behind the algorithm that recommends a product when you shop online or helps your bank detect fraudulent use of your credit card. Even the humanoid voice that announces the air temperature or the time of your next hairdresser’s appointment is just another piece of sophisticated software that involves many of hours of programming, cups of coffee, emptied beer cans, and sleepless nights. Nothing magical there. It’s just work–a lot of work.

Because this ant activity isn’t glamorous, because it doesn’t quench our thirst for sensationalism, the marketing departments of large AI companies have put their creativity to work. They have come up with truly inspirational stories, willingly relayed by a press that cares about circulation numbers. Those articles create a wow effect and nurture fears and hopes by turning AI into a social phenomenon. Even worse, everybody who has heard about AI seems to think they know what AI is.

The problem is that the image of AI conveyed to the public would fit more in a movie theater or a video game than in the business or technology columns of a newspaper.

AI is much more prosaic than what you’re led to believe. It’s algorithms that crunch gigantic amounts of data, find correlations, and make predictions. Terabytes of documents scrutinized, analyzed, and categorized. Highly secured data centers with endless rows of blinking servers. Really boring. Astonishingly, it’s exactly this boring stuff that helps improve our daily lives, not the miracles touted in press releases.

What we do at Cortical.io, for example, is nothing that would directly appeal to the media or the masses. We create intelligent systems that understand the meaning of text and help companies cope with the avalanche of text data they produce day after day. Creating an engine that processes millions of contracts and automatically extracts key information doesn’t sound sexy, does it? But it speeds up the authorization process of your new lease agreement. Employees don’t have to spend weeks reviewing a daunting pile of documents; the system does it automatically and rapidly for them. As an immediate consequence, the bank employees can put their intelligence to use elsewhere, focusing on higher-value, much more rewarding tasks.

That’s right: an intelligent system helps humans make better use of their time and effort. It doesn’t take away their jobs, it only helps them weed out the dull parts.

This means that the customer is happy, the employee is motivated, and the bank dramatically increases the efficiency of contract analysis, resulting in lower operating costs and higher profit margins. Nothing here to shout from the rooftops, but something that might have a profound economic and social impact as it spreads across the enterprise world.

Note that a Silicon Valley tech giant with an unlimited advertising budget and countless creative minds could transform such achievements in language understanding into impressive headlines. But they don’t. Why? Because, despite thousands of engineers and billions in research investment, they haven’t come up so far with an efficient solution that is deployable in an enterprise environment. Cortical.io, with its short history and a staff of about 20 employees, has.

But back to my point, it’s about time to demystify AI. AI isn’t coming; it’s already here. AI is not menacing us with some kind of supernatural power; it’s silently working day after day to help us live a better life. AI is not an ominous threat trying to escape our surveillance; it’s an accumulation of small, disparate systems that are programmed by humans to perform well-defined tasks–and just those.

AI as a buzzword has been so misused that we might as well consider it deprecated at this point. Maybe it is time to coin a new word?

Further readings:

Author: Marie-Pierre Garnier,
VP Marketing & Communications
back to opinions overview