The Eyes In The Sky Don’t Sleep
(Initially published January 7th, 2017)
Amazon is nailing it. After sweeping online commerce and redefining the world of cloud computing with Amazon Web Services, the company founded by Jeff Bezos does not stop experimenting with innovations, each more surprising than the last one.
First, it appears that Amazon is secretly building an Uber for truckers in Mobile App format that would connect truckers with cargo agents who handle freight. It makes a ton of sense: on the one hand technology is, at this point, quite trivial; on the other hand Amazon has a giant presence in the world of logistics and this type of service would allow them to disintermediate an 800 trillion dollar industry.
Second, we have Amazon Go, a more public project that we recently talked about in an article by Teknautas. With it Amazon intends to jump from online to offline commerce, joining both in a concept of supermarket 2.0 that has no cash registers, no queues and no hassle. Local cameras use advanced artificial intelligence algorithms to track your movements, record your purchases, and charge them to your Amazon account when you leave the premises. See and then believe:
This time the business as clear and juicy as in the previous case. The total sales of supermarkets in the USA are $ 649,087m, and it is not so much about perishable items as the fact that studies show how patterns of behavior are changing, fusing environments: searching online, in-store testing. Finding in the store, looking for reviews and alternatives online. The worlds are merging, and Amazon does not want to be left behind. However here the technology looks like science fiction movie. True?
Without denying the technical merit and the multiple innovations behind a system as well integrated as Amazon Go, the reality is that the field of machine vision systems guided by algorithms capable of adapting to changing patterns of behavior has decades of research behind and its products have been observing us already for some time, in places that would seem unsuspected.
Imagine the following. It’s four o’clock in the luxurious Bellagio of Las Vegas. Outside the streets are practically empty, but inside the casino, propitiated by constant music and lighting, there is still a lot of activity. You are working inside the surveillance center, an authentic bunker covered with screens on which converge images from the more than 1000 cameras hidden in the roofs of the rooms, the “eyes of the sky”. While watching blackjack tables, you notice something unusual. You can not describe well what, but there is something suspicious about that guy’s movements, the one with the cowboy hat: his hand has a curious position, and sometimes moves erratically along the edges of his shirt. After observing for about an hour, a gesture betrays him, and you can see how the fingers wander in the inner part of the sleeve. A call to the security of the lower floor and the ruffian is trapped with an ace in the sleeve, in a quite literal sense. Sam ‘Ace’, played by the brilliant Robert de Niro in the movie Casino, would be happy.
These are the scenarios that casinos have to deal with daily, and are a huge source of monetary loss. To prevent this, on the one hand, they have extremely qualified personnel as described above. But a single person monitoring a group of cameras is not a particularly effective system: people get bored easily, and keeping concentration across multiple monitors, gaming tables and clients can be difficult even in the best of circumstances. Not to mention the amount of time that is necessary to detect a particular pattern, a situation that is complicated when it is not an individual acting alone but an organized group to burst the bank counting letters that are coordinated by signals previously agreed. Hand in hair. Biting the corner of the lip. Rub the eyes.
On the other hand is the investment millionaire in software systems, as for example the database of recognized fraudsters that all the casinos share, and NORA (“Non-Obvious Relationship Detection “) that allows them to quickly determine if Players or dealers sometimes shared a phone number, a postal address, a hotel room … it may sound like science fiction, but it did not happen any more recently when it was discovered that by the corridors of the famous Venetian casino there were more falsehoods than the Water channel with half-day gondoliers and the roofs painted blue pastiche emulating the Italian sky: collusion between a dealer and a frequent customer to pluck the casino in front of his own nostrils jumped through the air when out into the light that the player Was included as a reference on the bank loan application form that the employee filled out when purchasing his previous car. It looks like a wild story coined by Danny in Ocean’s Eleven. But it’s real!
Using artificial intelligence, a system can go beyond the number of predetermined searches in the system, however large it will always be finite, and learn to find new exotic correlations. Likewise, security algorithms trained to monitor cameras can recognize complex patterns and alert the human operator if they perceive anything suspicious. This type of software was created for the gaming industry, but the technology has proven to be so effective that the US Department of Homeland Security has adopted it to discover connections between individuals suspected of terrorism, as well as being in the process of being used as intelligence. Business for banking processes, insurance companies, and a long etcetera. It should not be surprising, since after all, casinos have the money to employ cryptographers, security experts and programmers with more talent, even before government agencies.
The advances in the quality of both hardware and software make yesterday’s dreams the realities of today. The next time his erratic behavior by looking for the lavatories through the airport terminal with the nervousness of the few remaining ten minutes before the flight leaves you in front of two kind but firm security guards, Has been a human who has betrayed you, but has fallen into the range of statistical mistrust of an algorithmic system trained to identify the possible patterns of movement of a terrorist.
Do not get angry; Above all it is not personal.
