A memo to marketers: artificial intelligence is only as good as the engineers who created it.
So cautions Deirdre McGlashan, global chief digital officer at MediaCom, who chats with Ad Age on the floor of the Consumer Electronics Show in Las Vegas.
"Everyone is worried about the robot army coming and creating a doomsday scenario. But I think the real risks are [in not] understanding, when you're using products with AI in it, what was the data that was being fed?" she says. "How supervised was the learning? How broad and deep was that data set that it initially started with it, and the data set it now has access to? Without the correct depth and breadth, you have the risk of creating unconscious bias in these new machines and that can be very dangerous for us."
But McGlashan is no Chicken Little when it comes to AI. She couldn't be if she tried -- it's already permeated everything.
"We keep thinking it's this thing that's just around the corner. But it's already infiltrated our lives," she says. "It's in a lot of the parts of the infrastructure that we use that we don't even know about," from recommendation engines at online retailers to air traffic that guides our planes.
Still, despite its ubiquity, uncertainty hovers around AI — how to tell if it's "real" and what differentiates it from, say, a simple algorithm. (Artificial intelligence systems are often built out of a series of complex algorithms, which are themselves sets of instructions for a computer to follow.)
"Sometimes it's really hard to tell the difference between a good algo and AI," says McGlashan.
Marketers shopping for any given solution would do well to learn the difference, though only to a point. "Does it do what you want it to do? And are you being charged a price premium for the word AI?'," she asks. "Because if you are, then you do have to make sure it's AI. But if it does what you need and you're not being charged a price premium, then use it."