Today we shall talk about a polygraph also known as ‘a lie detector’.
To be more precise – about a European project iBorderCtrl.
Or if even more precise – about why the AI-polygraph functions in the new system of border control have an extremely low value.
And some last but not least – a spectacular example of how fake news spreads across the Internet.
A polygraph is a total of technical tools that measure the physiological reaction to stress caused by one or another issue. Its modern models can automatically record up to 50 parameters, even the unintentional lips trembling. But usually, it measures pulse, blood pressure, respiratory rate, and skin conductivity. All other parameters, such as changes in facial expressions and a voice timbre are noticed by a polygraph examiner who comes with the device in the kit.
First, the equipment is calibrated using test questions, the polygraph examiner knows the exact answers to which.
– Is your name Paul?
– Are you a man?
– Are you sitting on a chair?
Your parameters are fixed at rest, with both honest answers and false ones. As a result, there is an established personal relationship between physiological reactions and a certain level of stress. The procedure does not provide a 100% guarantee, and the result is affected by a subjective assessment of a polygraph examiner. But can this technology be widely used if the process is automated and is more reliable?
It may seem that by developing an iBorderCtrl the European Dynamics are chasing this very aim. The system is a set of hardware under the control of AI, which may check travelers at the EU border points. Why ‘may’? Let’s start with the beginning.
The system will work in two stages.
The first part of the procedure is remote. Before a trip, you will need to upload your passport, visa, and other necessary documents into the program. Afterwards, an animated avatar of an AI border guard is created, taking into account your gender, nationality, and native language. This is whom you have to chat with on the webcam. The system will give background information and ask a series of simple questions, mainly verifying your personal data. At the same time, this electronic border guard will monitor 38 unintentional micro-gestures and mimic reactions inherent in people trying to lie. In addition, the algorithm will collect available information about you on the Internet (for example, analyzing accounts in social networks) and compile your profile, including personal data, documents, photos, videos, and voice recording. And it is the creation of such a profile that is the key part of this news; the “lie detector” that recognizes your mimicry from a low-quality webcam picture is only a small auxiliary part, more like the distraction of attention.
According to the results of the first stage, each profile is assigned its own level of reliability (which the user, certainly, will not see) and a QR code which the border guards will use to get access to the profile. The assessment made by the AI will affect the thoroughness of control at the border: will they let you in after checking the documents or go over your baggage?
Currently, the system is taking laboratory tests, but in August 2019, it is planned to be tested at four border crossing points in Hungary, Greece, and Latvia. So far, the development has cost the EU about € 4.5 million. In the future, this will speed up the work of the border service and allow travelers to spend less time at the border, passing most of the checks online. What is more – it will provide the European Union with complete (including biometric data and information about activity in social networks) with profiles on everyone crossing the border. And this refers to 700 million people a year. What the effect of one or another parameter on the assessment profile will be is unknown. Of course, initially, the participation in the iBorderCtrl program will be voluntary. However, if the program is effective, it is obvious that one day the passing of such a control will become mandatory.
Another significant remark is a huge amount of fake news on the topic, which appeared on various resources. Again, basically all the ‘hot facts’ relate to a supporting system for evaluating facial expressions but not to profiling. For example, it is often said that the interview in front of a camera will take place right at the border. Obviously, the procedure of passing the control would not speed up. And some people thought that an animated avatar doesn’t sound truly cool, so the ‘digital’ border guards ‘will look like high-tech robots’. Agencies often cite tricky questions that will be asked by the system, watching your facial expressions: “What do you have in your suitcase?” and “If you open the suitcase and show me what’s inside, will it confirm that your answers are correct?” Not a bad idea, but how to deal with the fact that a user is taking the survey remotely, at home, and the suitcase might still not be packed?
We have managed to find out that the fake information about this technology began to spread from the article on newscientist.com, published on October 31.
In our piece, we primarily relied on the information from a study on the website of the European Commission, as well as on the official resource of developers. Here is their poster, which has the whole scheme of actions described in pictures. No robots in airports.
Want to receive accurate information? Keep on following our news.