Ir al contenido principal

Amazon Go: A reality we are not yet ready for?


For those of you that have never heard about it, Amazon Go is the lastest invention of the tech giant based in Seattle, an "smart" supermarket in which the client just enters using its smartphone, takes whatever he/she wants, and then leaves, without checking out.

And how does this whole revolutionary idea works? Well, the answer has just two words: ubiquitous computing. Yes, the shop is, mainly, a ubicom system itself. There are cameras everywhere in the place that tracks the clients, so they know what everyone is taking or leaving; there are also other kind of sensor everywhere, specially in the shelves, also to track what everyone is taking to buy. And everything is, in the end, connected to your smartphone, in which you will be charged for everything you buy.

Since this is, as we've said, an ubicom system, we can check that the 3 main properties proposed by Weiser are being satisfied:

  • Is the system networked, distributed and transparently accessible? Definitely yes, since every camera and sensor (which are distributed) is connected to the rest, probably also to a central control system (or more than one) and with your smartphone.
  • Is the computer interaction with humans hidden? Except for the check in the entrance in which you have to explicitly use your phone, the rest os the system never directly interacts with the user in any moment, not even for paying what everyone buys.
  • Is the system aware of the environment and the context? Well, that's the capital target for the whole system created by Amazon, knowing what everyone is doing in a not-so-much controlled environment.
As for the other two additional properties, we don't know for sure if the system is autonomous or not, although we can suppose that it is and not human intervention is needed when working correctly. And finally, regarding to system's artificial intelligence, we can't also be sure about that, but based on the information provided about the system, it probably doesn't have any visible intelligent behaviour, since it's just traking the clients and what they are doing. Nevertheless, is highly likely that Amazon uses some kind of AI-based algorithm for other purposes, as analyzing client's buying customs and likes.

Context-awareness problems?

As mentioned before, the context-aware system of the Amazon Go shop is mainly based on ceiling cameras and a set of sensors in the shelves that controls what he clients are buying i.e. what they take from the shelves. Taking this into account, we could say that is a physical environment context-awareness system, not controlling either the human context (has far as we know) or the virtual context. It's also an active system, since everything you buy is automatically added to a virtual shopping cart that you will be automatically charged when getting out of the place.

Nevertheless, this system has been a headache for Amazon's developers. In the first place, the opening of the shop was planned for several months before the actual one happened. It's rumored (if not confimed) that the cause of that delay was that the contex-awareness system crashed when there were more than twenty people inside, since the traking system was unnable to follow everyone.

Now that they have apparently solved that problem, they are having some other problems that are also quite annoying, not just for them but also for the clients. Sometimes, the cameras happen to confound people with similar body shapes, which is obviously a problem since they could charge them with items they haven't bought. Also, they have reported some other problemas with the sensors on the shelves, mainly based on that some people (specially children) are moving or misplacing items to other shelves, so sensors are completely lost with these new objects placed on them.

Are we ready for this technology?

One of the thigns that most people that have visited the new Amazon shop stated is that they could buy a lot of stuff in just a few seconds or minutes, almost without noticing and expending a considerably higher amount of money than usual. Some clients reported to have spent 50$ in less than a minute, almost without being conscious of it. You just take what you want, and get out, without having to pass through the checkout and having to explicitly pay for what you have bought.

Another problem derived from this checkout-less way of buying is losing the contact with humans, leaving everything to computers. Following this, some people started to post some comments on the social networks stating that if this way of buying would spread to every single supermarket, they would just miss the checkout line and having a little chat, an interaction with another person, when going to pay.

Finally, related with the contetx-awareness problem that has been mentioned, some customers that visited the shop also stated that they were feeling like if they were stealing from the supermarket, going out without paying, which in some cases was true, since the traking system wasn't working correctly and it wasn't charging the clients for everything they have bought, so some people found they were kind of shoplifting when they notice some of the products weren't on the ticket.

Privacy issues

As Barry Steinhardt said on 2000, "Soon you won't only be able to capture how many people stopped by, but who they were.", and in the case of Amazon Go, this sentence couldn't be more accurate. While this revolutionary way of shopping definetly reduces shopping time, some people sees it as another way of surveillance that could have gone too far.

Again, and following his previous sentence, Steinhardt added that "Once identity is established, it will be cross-referenced to capture that person's income and buying preferences.". As mentioned before on this entry, we still don't know if Amazon is using any kind of AI and data mining technologies to get that kind of information about everyone that enters their shop.

As you may know, there's been recently a major leak on Facebook that has compromised a lot of private information about its users. So the obvious thinking now could be: what if Amazon also suffers from an information leak? Would they compromise not only our names, addresses and other private information, but also our bank and credit card numbers? Our buying preferences? So far, we are just relying on them for that not to happen.

The future of Amazon Go and ubicom technologies

Althoug all this would look like as a lot of problems and a loss of money for Amazon, the company has recently announce that they will be expanding their shops, along with its revolutionary technology, to six new locations across the US by the end of 2018. We may or may not be ready for the future, but it's comming, whether we like it or not.

Comentarios

Entradas populares de este blog

"Oh my, I didn't want to do that!" - Preventing User Errors in Interaction Design

Who hasn't been in this situation when you are using a smartphone app or a webapp on the PC and you make something completely wrong, sometimes even with no option of going back? If someone says that he or she doesn't, either he/she is lying or he/she has never used an app. But, who's responsible for this error? The easiest answer (specially for developers) it's always "the user". But nevertheless, if the user is doing something wrong, it's because there's something wrong or that can be improved in the app. There exists two different so called "user errors": slips and mistakes. Slips Specially nowadays that everyone has one or more smarphones, and that we even use them more that a PC, slips tend to appear in everyone's daily routine. For example, you have an appointment next week and you don't want to forget about it, so you program a notificaion on the smarphone's calendar; but when you are going to save the appointment...