Watch this episode on YouTube.
Reasonably Accurate 馃馃 Transcript
Morning everybody. How are you doing today? In this episode of the show? We're gonna tackle facial recognition again. Now I say again because we've talked about it before and it's an issue that kind of keeps coming up again and again. And there was actually three media articles uh in the last couple of days that really triggered an interesting viewpoint uh for me or sort of line of thought around facial recognition.
Now, the first comes uh from uh zero day which is a column or subsection in ZD net and they were talking about how the E decided to create a giant biometrics database. Um Now that seems initially horrifying um one central place with everybody's biometrics, whether those be fingers or retinal scans or facial geometries.
But then they go on to uh examine the primary use case, which is to help facilitate border crossings. So keep that one in the, in your mind in the back of your head as we kind of go through this, the second article was from the Financial Times. So it's up on ft.com.
And as always with all the things that I mentioned here, I'm going to post those links in the description down below. So you can reach the original source material that kicked off this line of thinking. So the article on ft.com talks about how the training databases used for really popular facial recognition tools and services, the people in those facial recognition databases might not know they're there.
Now, at first you're thinking, wow, they're using their identities without permission. But one of the things that popped up in that article was that um for one particular data set at the very least was it was using Creative Commons attributed or licensed um photos, which means that they probably had the legal right to use those photos.
The question is whether they the moral or ethical, right? Um So coming back to that one in a second again and then the last article was from the hill talking about how uh DH S in the States wants to get um near 100%. I think it was like 97% coverage of uh using facial recognition for people exiting the country um through airports.
So three interesting articles again, the links down below. So you can read them yourself. Um But all centering around using facial recognition, which is probably the most common biometric use right now. Um You know, for general purposes outside of straight authentication uh on devices which then probably thumbprints don't have any hard stats around that.
But I think facial recognition raises the biggest um community or societal discussion because I can apply facial recognition based on an image based on video and from a distance for thumbprint, you know, you need to actually put it up there and scan it. Um So there's a little different interaction case here.
Now, the EU wants to help facilitate border crossings. DH S wants to track people leaving the country. And these training algorithms were using photos that were probably already licensed for their type of use. And this is where it gets really interesting because the crossing the border, you're already being identified and you're already heavily recorded.
So combining those two doesn't seem like it's such a leap, but there's something sort of viscerally that goes in. I'm not really comfortable with that same with the border case for the airplane case, the airport terminal case with DH S. You're already heavily monitored and photographed within the airport terminals.
You've already provided identification in order to pass through security or to get to the gate, depending on the country you're in, you've already passed these stuff, you already um consented to this. So again, merging those two makes sense. Um And for the training databases, if they've at least adhered to the licenses of the photographs, you've already consented in a way.
But all three of these feel kind of off and that's the best way I can describe it is that they feel kind of off. And the reason for that is because we haven't discussed those specific use cases, the exist um or the consent that has been given in these cases has not extended to the specific implementation.
And I think that makes a world of difference. So yes, you said, you know, you can use this photo to reprint or to attribute um under a creative commons license or to spin it off into something new. But nobody thought about the use case of that being used to train algorithms and services that could be used in a manner that you're not comfortable with.
You've already consented by traveling across certain checkpoints and borders to being identified and to being recorded within the terminal for security reasons, right? Um And for a whole bunch of additional reasons, which is why I put up the air quotes. Um but you've already consented to that and you're already comfortable or at least accepted that.
But combining the two to be able to do it at speed makes people uncomfortable because it's not something we've explicitly discussed. And really the point of this show today was to bring that up is that facial recognition is being pushed more and more into the spotlight.
And I think the underlying theme in every single time this pops up in the headlines is that we have to have a discussion in each community, in each society. We need to discuss how we're comfortable with this technology being used and what scenarios and what circumstances and this is going to be a case by case, there could be broad guidelines, but there needs to be a case by case discussion around this because maybe you're comfortable with it at the airport.
But are you comfortable with it in the mall? Right. And we've seen that here in Canada where apparently some of the mall kiosks that you could help uh touch to get uh directory, store directory, see what these sales are on that kind of stuff. Apparently, some of those were doing generic facial recognition, they weren't identifying, you know, this is Mark's face, but they were saying this is a unfortunately middle age, uh graying white caucasian.
Um, you know, and, and they could use that for demographic information is that most likely I'm looking for stores that align to those interests as opposed to, um, you know, maybe other stores like, uh you know, that might not align with that particular demographic. So again, it's context it's use in context.
And I think the longer we go without having a discussion, the more that the horse is already going to have left the barn. Um It's going to be really hard to put this back into play because people have already deployed the technology and have been using it in the argument will be, well, it's been in use for years.
So what's the difference now? Um And we're gonna see some uses that we're really not comfortable with. And we've seen that with a pushback from some of the technologists who work on this and the researchers who work on these technologies. But I think this discussion needs to come out into the light.
And that's really the goal of today's video is to make that clear to everybody and hopefully everyone who's watching this can help start that discussion um so that we can have this and decide where we're comfortable with as a community and that's going to be different on a community led community basis.
What do you think? Let me know, hit me up online at Mark NC A in the comments down below. And as always by email me at Mark N dot ca, I always find it fascinating discussing these and other issues uh with you folks. It's really a community driven show.
So as always, if you have ideas, let me know, happy to do a video on any topic moving forward because just like this su subject around facial recognition, the more we talk about this stuff outside uh in the light and work through these uh really tricky edge cases.
I think the better off we all are have a fantastic day and we'll see you on the next show.
