Artificial Intelligence And Autonomous Cars. A real Threat To Humans?
Autonomous Cars, every morning we get up, and in our day to day many, we could say even too many, smart devices accompany and surround us: watches, televisions, refrigerators, security systems, vacuum cleaners and even toys. A list that is constantly growing and seems ever closer to infinity, even making us think that the only thing that is stopping to work intelligently is the human being.
Although we could say that Artificial Intelligence (AI) is a relatively recent term, since a little more than 60 years ago, by the hand of computer scientist John McCarthy, this term was coined for the first time during the Dartmouth conference in 1956; the truth is that its progress has been unstoppable for some time.
Although the attempt does not stop, and great achievements have been made in the field of AI: machines have been taught to see, read, listen, speak or even drive, it seems that none of this can do it the way we humans get it, right?
Precisely this last skill, autonomous driving and connected cars are reaching a degree of evolution that increasingly borders on human driving.
Currently, vehicles that have cruising speed or assisted parking systems are already available on the market and are easily accessible, but for years, the automotive sector has been in continuous development and progress, presenting itself as imminent coexistence on the roads of traditional and autonomous vehicles.
In short, we will have smart roads hand in hand with 5G technology, that is, an infrastructure that is also connected.
All of this will mean that humans, as mere passengers, will travel and move around in a fully connected environment.
The advantages of the roads and transport of the “future” are more than evident: safety and reduction of accidents, roads more open and connected with nature, comfort or a cleaner environment. However, we must also pay attention to the other side of the coin: the disadvantages, which cannot be considered few. Thus, I present below some of the disadvantages or negative points that such highly developed technology carries with it.
Mass Storage Of Information.
It is nothing new to say that smart or connected cars present significant challenges regarding privacy and the protection of personal data. Knowing the route on the way to work or our home, the contact list or our voice or data that allow us to know guidelines or lifestyles, is configured as information that is increasingly accessible for car companies, thanks to the use of applications such Audi Connect, Mercedes Me, Apple Carplay or Android Auto, among many others.
The use and storage of all the information that we generate when we use a connected vehicle are subject to current regulations on data protection, that is, the European Data Protection Regulation (RGPD), and Organic Law 3/2018 of 5 December, Protection of Personal Data and Guarantee of digital rights (LOPDGDD), as long as personal data is being processed.
This will mean the need for, among many other obligations established by the regulations mentioned above, manufacturers or entities that may intervene in the processing of personal data, provide clear and transparent information to citizens, which in any case complies with the requirements of articles 13 and 14 of the RGPD and 11 of the LOPDGDD. In this regard, and although the essence of this regulation is precisely to grant greater guarantees and protect the rights and freedoms of the interested parties, it is still recommended and almost necessary being aware of the general tendency to breach the regulations or compliance with minimums, as users of these services we pay special attention to the terms and Privacy Policies of the companies responsible for the Apps that are responsible for connecting our vehicles with the world.
Autonomous Cars As An Instrument Of “Attack”.
Suppose almost two years ago the fatal accident caused by an autonomous car of the well-known transport company Uber made headlines, highlighting the technical failures or the risks that the human losing control of the machine can pose. In that case, it is time to put on the table how “tempting” these vehicles can be for any computer attacker.
The fact of turning cars into an intelligent and connected machine means transforming them into a gateway to possible cyber attacks that can not only put the information stored in the vehicle at risk but an even more precious asset than our data: ourselves or any other person who crosses the path of an autonomous car, in short, human lives.
Let’s think that autonomous cars act according to the programming carried out. This implies that, for example, by having vision and information processing systems that allow the identification of pedestrians, it is as simple as modifying a few pixels of those images and eliminating a pedestrian, thus avoiding the braking of the vehicle and consequently trample anyone who crosses his path.
Can The IT Security Of These Vehicles Be Improved?
Being aware that, once again, zero risk does not exist, it has been shown by mathematical specialists that it is possible to solve or reduce these challenges. There is precisely a European project ” Trustonomy Building Acceptance and Trust in Autonomous Mobility ” (build acceptance and confidence in autonomous mobility), which aims to improve driver less cars over the next three years, with the participation of researchers from nine countries.
For its part, the National Cyber security Institute (INCIBE) has also spoken out in this regard, and in a publication related to smart cars, in which it analyzes the communication protocols and technologies used by these “vehicles of the future”, also It includes what the attack vectors are in smart cars and how to avoid them.
- Manipulate the information of the ECU (Engine Control Unit). The attacker could be able to extract and analyze information from the ECU, which could lead to an alteration in the correct operation of the vehicle.
- Proposed measure. Prevent physical access to attackers by controlling who has access to the vehicle.
- Bluetooth attacks. They allow, for example, the theft of sensitive data from the person who makes a pairing of their device.
- Proposed measure. Disconnect Bluetooth whenever it is not necessary and keep the devices updated with the latest version available.
- Use of malicious applications – Applications that can be found and downloaded from untrusted repositories.
- Proposed measure. Avoid using applications from unknown developers, always download from trusted repositories, such as the Google Play Store or the App Store, or from the official website of the manufacturer.
- RF attacks. Depending on the model of the vehicle and depending on the algorithm you have when generating the keys, these could be stolen and later used to open, close or start the car.
- Proposed measure. Automatic generation of keys with encryption each time the key is used, thus preventing the same keys from being reused.
In short, the attacks that manufacturers have to include: alteration of the measurements made, generation of errors or codes that the car interprets as dangerous for its operation; or even more serious, as researchers have shown in POC (Proof of Concept or Proof of Concept) by braking a car, or achieving absolute control of the vehicle, as has been demonstrated by researcher Charlie Miller.
Ethics And Morality In A Machine, Is It Possible?
Notwithstanding all of the above, and despite the fact that there are technical issues such as perception, which pose a challenge for autonomous vehicles, in this case, the greatest obstacle for the machine to overcome or equal the human being, is intuition. The capacity of the human being that has a direct impact on the moral decisions that in certain situations, must be taken behind the wheel.
As I mentioned in this publication, autonomous cars will act according to the programming carried out, and this leads us to ask questions directly related to the famous Tram Dilemma: how would an autonomous vehicle act in an unforeseen dangerous situation? Would you sacrifice the occupants of the vehicle, the occupants of another vehicle, or the pedestrians on the sidewalk?
Although the Mercedes company seemed to have it clear in 2016 when it declared that they would choose in favor of the passenger over the pedestrian, whoever it is, the answer to these questions is not simple. Why? At the end of the day, it becomes clear that the autonomous car must be prepared to assess difficult ethical situations, and therefore we are talking about problems that are beyond the scope of mathematics and engineering. That is why, in the aforementioned European Trustonomy Building Acceptance and Trust in Autonomous Mobility project, philosophers, among other professionals, also participate, reflecting on these ethical issues.
In short, and taking into account what is stated in this publication, although there have been gigantic advances that years ago were considered unthinkable, we are far from fulfilling the forecasts that four years ago placed 10 million cars in the current year 2020 freelancers circulating worldwide. It thus shows once again that there are still some obstacles to overcome before we put ourselves in the hands of autonomous vehicles within a connected infrastructure, and we will maintain, for the moment, control over our vehicles.