Accident puts the safety of autonomous cars in doubt

Could an autonomous car accident have been avoided? Running over of woman raises doubts about control and supervision. It seems that the difficulties for the autonomous car are just beginning.

An autonomous car, which was part of Uber's testing program, ran over and killed a woman, in the city of Tempe, state of Arizona (USA). This was an accident that hit the reputation of autonomous car technology, which does not require the driver to direct.

One of Uber's autonomous cars, which is part of the company's testing program, ran over and killed a woman last Sunday night (18), in the city of Tempe, Arizona (USA). The accident marks the first fatality involving a car with autonomous driving technology, which eliminates the need for the driver to drive.

An Uber employee was inside the vehicle, because during this testing period there is a driver to supervise the vehicle's performance. The car involved in the fatality was a Volvo SUV model, the XC90. The collision occurred when the victim was crossing the road, holding his bicycle. The front part of the car was badly damaged.

Image: Folha de São Paulo

Uber publicly apologized for what happened and cooperated with local police. The company announced that it will suspend testing in the United States.A video was recorded by the vehicle's own camera and shows in more detail the fatal accident, which was the first recorded with this type of car. The vehicle was traveling in autonomous mode and the video shows that the operator, who was inside the vehicle, looks down for five seconds and then looks back at the road, practically at the moment the car hits the woman. The victim, who was holding his bicycle and crossing a poorly lit street on foot, was identified as Elaine Herzberg, 49 years old. The video shows that the driver could see the woman's feet, illuminated by the car's headlights. The images were released by the police.

Tests with autonomous vehicles have been suspended

In the week following the accident, Uber suspended tests with autonomous vehicles, at least temporarily. Tests were being carried out in Pittsburgh, Toronto and San Francisco.

The police stated that, possibly, the error for the accident could not be attributed to Uber, because the woman who was run over was not crossing the road in the lane. Perhaps, according to the police, the collision could not be avoided, because it was a dark street, even if the car was not autonomous.

Following Uber's temporary suspension of its testing, authorities in the state of Arizona (USA) banned Uber's testing of autonomous vehicles in the state. Arizona Governor Doug Ducey called the released video footage of the crash disturbing and alarming. According to him, doubts were raised regarding Uber's ability to continue with its tests and announced the suspension of the autonomous car program in the state.

Uber autonomous car kills woman in the United States

Accident marks the first fatality involving a technological vehicle

A new problem for technology

After many problems with bans in some cities, dissatisfaction among its drivers and reports of harassment by some users, the news of the accident with its autonomous car being tested further damaged Uber's image.

In 2017, the company was accused of prejudice against its employees and faced accusations of industrial espionage against Waymo, a subsidiary of Alphabet, a Google company, for allegedly stealing the development plans for its autonomous cars.

Legislative problems occurred in several countries, including Brazil. The former president was the controversial Travis Kalanick, who was replaced by Dara Khosrowshahi. The company's image crisis was managed by Bozoma Saint John, who was head of marketing for iTunes and Apple Music. News about its finances revealed losses, with a growth of 61% in 2017, compared to 2016.

Who is to blame for the accident

The accident in Arizona happened two weeks after Governor Doug Ducey signed an executive order that made fully autonomous cars official on public roads in that state. The accident puts the government on the receiving end of public scrutiny of its decision.

The governor's order established that autonomous cars must obey traffic laws and all existing rules. According to the government, the advancement of technology requires that policies and priorities adapt to remain competitive in the economy. The state of Arizona is an operations center for autonomous vehicles and there are already 600 cars in operation, from Intel, Waymo, GM and Uber.

Source: Oficinadanet

Arizona and California have approved driverless cars, even without supervisors at the wheel. Tests with 100% autonomous vehicles would begin in April. Legislation is adapting to technological innovations, but legally the question remains: who is to blame in an accident in which artificial intelligence kills?

On this subject, Professor John Kingston, from the University of Brighton (UK) published a scientific article at the end of February, suggesting the possibility of an autonomous car killing someone, which actually came to pass. According to the professor, who used as a reference the work of researcher in Criminal Law and Artificial Intelligence Gabriel Hallevy, from the ONO University, in Israel, determining the guilt of a crime necessarily requires action and intention.

Image: Rota42

Based on this statement, a crime committed by an artificial intelligence system could occur in three hypotheses:

1 – The culprits could be the artificial intelligence programmer or whoever uses it. It is the type of crime “perpetrated by someone else”.

2 – The event could be of probable natural consequence, which happens when the program performs actions inappropriately, which is the reason for the crime. As an example, a robot may consider that a human person is getting in the way of carrying out its task and may therefore attack the person. This action could be considered “natural” for this type of artificial intelligence.

3 – The crime is committed through direct guilt, which would require an intention and action from artificial intelligence, something like an autonomous decision to attack and kill, which we have already seen in science fiction films. In the case of the car, which exceeds the speed limits, the blame could be attributed to the program and the programmers or users could be exonerated.

In the case of the Uber autonomous car accident, it seems that there will be a difficult legal problem to resolve. There was a driver at the wheel and his role was to take over the steering in case of danger. Who will be held responsible now, before the victim’s family and the State? It seems that the accident will raise further reflections on the production, inspection and operation of this type of technology.