05 May The ‘Deep Fake’ threats online – Joel Leslie
The future of A.I. is both amazing and frightening. It can help our lives in so many useful ways and things just get easier, like never having to worry about running out of supplies at work or home, organizing appointments or keeping you safe in your home and in the community. But what about the A.I. we aren’t considering, like Deep Fakes. Online guru Joel Leslie explains the dangers ahead. This will shock you!
Kevin: The future of AI, artificial intelligence, is both amazing and frightening. It can help our lives in so many useful ways and things will just get easier, like never having to worry about running out of things at work or at home, and even organising appointments. But what about the AI that we aren’t considering, like deep fakes? These aren’t just about copying the identity of a celebrity or a head of state. It goes, actually, far beyond that. And we’re teaching AI now to get smarter, so is that increasing the danger?
Kevin: Joel Leslie, from Property TV and expert in online technology development has attended the Inman Connect conference in New York earlier this year, and he joins me to talk about this topic. Joel, thanks for your time.
Joel: Thanks, Kevin. Thanks for having me.
Kevin: Tell me about the conference in New York. What did you notice is one of the biggest AI growth areas?
Joel: Look, AI is a hugely growing industry right now, not just in property. But property seems to have really taken the ball and run with it.
Joel: One of the great achievements I saw this year … quite a few companies too were actually using AI to detect what is in an image. And that is huge impact into the real estate industry right now.
Kevin: Explain to me why.
Joel: So if you consider, when you put an image into a classifieds website, like a search results, they’re generally considered a flat image but they’re actually full of data. So if you consider a kitchen that’s in someone’s house and you’re putting that onto a classifieds website, imagine if you could tell that they … a certain type of bench top or a Smeg appliance, and the width between the window, how wide the fridge is. All those things are actually in the image, and it just takes a little bit of nous to work out how to get it out.
Kevin: Okay, so once it comes out is this helping companies target me if I like a photo or if I put a photo up about a property?
Joel: Well, both. So you can actually think of two different markets here. So not only now do I not have to describe the image that I’m loading into the real estate website for those consumers to search, the actual AI will say, “Hey, there’s a kitchen here. There’s a Smeg appliance here. There’s a granite bench top here.” So not only do I not need to describe that anymore, I can also have those images now that are detected and then advertising extracted out of that image.
Joel: So if I’m one of the big real estate portals of the world and I have advertisers that are external to the property industry, let’s say home improvement or appliances, what better positioning for those advertisements on those real estate websites than to be beside an image that actually displays how they look in an actual home setting?
Kevin: So this is all about teaching AI to get smarter. Let’s look at some of the downsides of this. As we, obviously, teach it to recognise this stuff, how easy is it for it to make something fake?
Joel: Very, very easy. And what’s happening at the moment, there’s this concept … well, not a concept, it’s actually something that’s happening right now, and it’s called deep fakes. And what they are is it’s actually AI that’s learning from this imagery, learning from the things that make up a photo, but then assembling them into a brand new photo. And so one of the most recent uses of these, or one of the most known uses of these in the last 18 months is actually duplicating leaders, so let’s say Donald Trump or whoever it might be in other parts of the world, talking and saying things that aren’t actually them saying it. It’s actually a robot, and written a script from someone that isn’t necessarily saying that. And it actually goes far beyond that. I think what you’re touching on here is how does that impact the real estate industry.
Kevin: Right, it’s going to be my next question.
Joel: Yeah. So it actually does that too. So imagine you could emulate a real estate agent that didn’t exist. They’re vouching for this property that didn’t exist. But what about if the property never existed? And so what’s actually happened at the moment is there’s some very clever tech out there that’s actually putting together real estate listings that are for sale that just don’t exist.
Kevin: So I guess the real danger there would be if you’re buying without physically inspecting the property, and I know some people do that. That could lead to you actually losing a lot of money.
Joel: Exactly. And like you said, that’s common. People do do that. So specifically investment, there are investors out there who do invest in properties that aren’t inspected.
Joel: But also in the holiday letting scene, the long term and short terms, you can rent a property sight unseen and you can lease a property sight unseen. And that happens all the time, if you think about real estate websites, specifically TripAdvisor or whoever it might … Flip Key I think use their version of that. Those types of properties you turn up to in San Francisco, you’ve booked this property from Australia to go and stay at, and it never existed. And I’m not saying that they’re copying photos from the internet and putting it into a listing. These photos of the lounge room, of the kitchen never, ever existed. Everything is placed in there from the furniture to the view to even the person standing beside the bed.
Kevin: There are a couple of examples … I should have mentioned at the outset, too, that Joel has kindly written a blog article for us which you can get to on Property TV. We’ll also put it on Real Estate Talk as well. But if you go into the transcript below this interview, I’ll make sure that we put a link in there that will take you to this article because there’s some really good demonstrations of how this works, and you’ll see how effective it is.
Joel: Put it in there because there’s some very good links in there that’ll actually take you through to see how it’s actually being used right now. One of the most frightening things you’ll see is that the photos of the people that are put together who look real have never, ever been born and do not exist.
Kevin: Okay, well what technology is coming through … I think I know the answer to this one … that could actually help with this, giving us more security online?
Joel: There’s a lot of work being done at the moment on to validate, and I think that’s the big key here, validate and trust this type of technology. And it all comes down to who you know and how well you know them. Even within this environment that you can have someone that you’re talking to online that has never breathed in air on this planet, this is where technologies like blockchain will come in and really help this solution, because it is a trusted solution that can never be altered. So it’s back to the certification, it’s back to the trust, and it’s back to the verification of the person or the entity that’s putting the listing online.
Kevin: Okay, well, check it out for yourself in the transcript below this interview. You’ll see that link that’ll take you straight through to Property TV. You can look at the article. You can also click on the links and just see exactly what Joel is talking about.
Kevin: Joel, thank you so much for your time and for drawing our attention to this. I guess the bottom line is really just be aware that this could be happening.
Joel: Absolutely. Just have a look, be out there and do your investigations first, and know who you’re dealing with.
Kevin: Joel, thanks for your time.
Joel: No trouble. Thanks, Kevin, appreciate it.