< An Eye For An Eye

Transcript

Friday, February 06, 2009

BOB GARFIELD: How do you know when the nightmarish Philip K. Dick’s sci-fi future is upon us? When you look at a billboard and it looks back at you. You may remember that scene from the film Minority Report. Tom Cruise’s character John Anderton walks into a store and is bombarded with ads that call him by name.

[CLIP]:

[MUSIC UP AND UNDER]

WOMAN: The road you’re on, John Anderton, is the road less traveled. Make sure -

MAN: Good evening, John Anderton.

MAN: You’re here, John Anderton.

[END CLIP] [OFF-MIKE COMMENTS] Are we approaching the day when our billboards know us well enough to pitch their messages especially to us? Not yet, but maybe soon. Paolo Prandoni is chief scientific officer for the software company Quividi and he’s helped engineer real-life billboards with the ability to know and take note of who checks them out. Paolo, welcome to On the Media.

PAOLO PRANDONI: Thank you.

BOB GARFIELD: So I'm at the mall and I see a billboard there or some sort of sign, and what happens next?

PAOLO PRANDONI: Basically our system is a souped-up turnstile that counts the number of people that are actually looking at the billboard and provides some extra qualifying information such as your gender and the amount of time that you spend looking at the contents of the billboard.

BOB GARFIELD: My gender. Now, I'm not leaving any DNA behind. How does it know?

PAOLO PRANDONI: Well, it looks at your face, and according to your features it can formulate a pretty educated guess as to whether you’re male or female. So, for instance, if you have a beard, well, it pretty much gives it away. And other traits are more subtle, but it’s a combination of different hints in a certain direction. So, for instance, long hair would tilt the system towards female gender, full lips as well, or high cheekbones, for instance. There are small hints that can be combined into a pretty cogent guess.

BOB GARFIELD: Long hair, full lips, high cheekbones. So a Fabio walks by, you’re going to count him as a woman.

PAOLO PRANDONI: It’s likely, yes.

[LAUGHTER] I don't know if it’s a problem for Fabio, but it’s a possibility.

BOB GARFIELD: [LAUGHS] But I understand that you can also do age.

PAOLO PRANDONI: Age is a much more slippery quantity to assess because, of course, aging patterns are very different. So what we do is we try to define some broad categories where we differentiate, of course, between children and adults, and then we further split adults into younger adults and older adults, and then senior. And, again, the way we do that is by looking at features that somehow give away the age of a person, but we can't really attach a precise numeric value.

BOB GARFIELD: Okay. Now, like many people, I don't like to have cameras trained on me when I'm not aware of it. Shouldn't I be alarmed that I can't even walk past a billboard without fear that a camera buried inside of it is scrutinizing my image?

PAOLO PRANDONI: Oh, dear. So that’s a very, very [LAUGHS] controversial point. We don't record an image, and you’re nothing more than just a head count with a little tag of attributes that are attached to it, such as your gender and your attention time. I think there’s a little bit of technophobia that is attached to cameras. People are afraid of being scrutinized and they're afraid of being read by an evil marketing system. But the truth of the matter is that the system is not that sophisticated and, again, people are just head counts.

BOB GARFIELD: What if the police come in with a search warrant and they say, oh, by the way, we need this for surveillance for one reason or another? Do you have the capacity to accommodate them?

PAOLO PRANDONI: No, we don't record anything. And I think the police would be much better off to have a system designed for video surveillance and maybe get a better view, because a screen in the mall is not particularly big. The longer answer is, wherever there is a camera there is a possibility for recording, but that’s not what we do.

BOB GARFIELD: Does the content of the advertisement change based on who you perceive to be watching it?

PAOLO PRANDONI: Right now the gender counting is used mostly as a post facto analysis, but that’s probably going to happen in the future. Instead of being bombarded by ads that do not really apply to us, we would move to a situation pretty much the same way that it happens on the Internet, where the little ads that we get on, say, gmail are not that irritating because they talk to us more than the useless droning of messages that do not really apply to us. I mean, advertisement is not 100-percent evil. There is an informative component of advertisement that we actually are very happy to have in our lives.

BOB GARFIELD: Bearing in mind that this conversation is being recorded, would you, as the chief scientific officer of Quividi, care to tell me what percentage of advertising is evil? -

[LAUGHTER] - because I'm dying to hear that number.

PAOLO PRANDONI: [LAUGHS] Well, in my opinion, for instance, an advertisement that targets children is evil. And I am not against using my technology to switch off the billboard when the audience is composed of children, for instance, and that would be a great service.

BOB GARFIELD: All right, well, thank you very much, Paolo. I appreciate your talking to us.

PAOLO PRANDONI: Well, thank you. It was a pleasure.

[MUSIC UP AND UNDER]

BOB GARFIELD: Paolo Prandoni is the chief scientific officer for Quividi, a French software company.