Barriers to the Age of Autonomous Ridesharing

At the 2022 Consumer Electronics Show in January, Mary Barra, CEO of General Motors, announced that the company’s autonomous vehicles will be commercialized as soon as the mid-century (2025). On February 28th, The California Public Utilities Commission (CPUC) permitted Cruise (a subsidiary of General Motors) and Waymo (a subsidiary of Alphabet Inc) to provide autonomous shared rides service with a safety driver present. The future of autonomous vehicles is becoming very real and manifest in our daily lives.

In the age of autonomous vehicles, the business model of both vehicle manufacturers and ridesharing companies will change substantially. Individuals are no longer the primary consumers. Ridesharing companies would be the buyers and play the partner's sides in terms of technology investment, as in the case of Uber and Toyota. Any technology failure or malfunction could lead Share-riding to complicated legal charges by customers, which are costly and brand deterioration. To succeed in the Age of Autonomous ridesharing, ridesharing companies must overcome the following challenges.

Functional Challenges

Imperfect technology: Currently, autonomous vehicles are made with sensors to analyze the surrounding environment and make driving decisions based on a wide range of predictions. Since driving experience is complex, there is not enough data or collected scenarios simulation as Waymo did to train perfect AI without having the probability of malfunctions. Ziyuan Zhong, a research assistant in autonomous systems at Columbia, pointed out that simulation environments that companies use to train AI could lead to a dead-spiral effect. A dangerous real-life situation could be interpreted as a normal condition in a simulation environment. Think about autonomous car rides at the limit speed of 45 miles per hour as simulation, but under heavy snowy storms in real-life. In terms of advanced driver assistance systems (ADAS), autonomous vehicles use two major technology applications to “see” the surrounding, namely: optical cameras + radars (Tesla), and LiDAR – Light Detection and Ranging (Waymo, Uber, Yandex, and Toyota). Still, none of these technologies prove the perfect solution to automated vehicles. Optical cameras + radars could be blind by extreme light (sunlight) and don’t picture the depth and range of the information. While LiDAR is a safer and more reliable alternative, the technology is expensive and could be tricked with a handheld laser. Ridesharing companies must find the sweet spot to balance between technology and the cost of service.

Optical cameras + radars LiDAR ( Light Detection and Ranging)

Cyber-attacks: Applying AI in autonomous vehicles requires ridesharing companies to develop various methods to protect their cars from undetected threats that could cost the company millions of dollars, including car theft, overwrite control, system manipulation. At Blackberry Security Conference in London, Mr. John Chen, CEO of Blackberry, said that self-driving cars have more complex code than a conventional combat drone, which means hackers have many “gateways” to attack and take control of the vehicles. Hackers could overwrite systems to mislead fare calculation, alter payment information, or turn the car into a dangerous weapon for kidnapping and extorting passengers. On Jan 11th, 2022, a 19-year-old German citizen claimed he successfully hacked into more than 25 Tesla cars in 13 countries without drivers' acknowledgment. In November 2020, Lennert Wouters, a security researcher at Belgian university KU Leuven, could also create a $300 hardware kit that could hack into Tesla Model X within 90 seconds.

Ethical Challenges

Customer safety: customers don’t have control over the vehicle, so they can’t react or take control of the car in unwanted situations, namely: natural disasters, unexpected weather, or AI malfunction. Since 2018, out of 234 Tesla deaths, 12 cases have been related to Tesla Autopilot issues. There is a thin light between the cause of vehicle accidents, which could attribute to manufacturer errors or operation errors (Ridesharing companies). At the same time, there are limited legislations on tracing back and defining the responsibility of each party if these situations happen. However, as service providers, Ride-sharing companies would be the first party to be claimed if accidents happened. To mitigate the risk of malfunctions and legal charges, these companies must invest in costly maintenance infrastructure while developing legal guidance to deal with legal threats. Share-riding companies and autonomous vehicle manufacturers would also experience the potential conflict of safety priorities. While share-riding companies fight for passengers' safety, autonomous vehicle manufacturers must balance the passenger and outside entities' safety. There is still a big question mark for what AI behavior would be when it comes to these rare situations when it has to choose between third parties’ life or passengers’ life, which required situation analysis and the sense of empathy, tolerance, and sacrifice as a human being.

Customer privacy: As autonomous vehicles become popular, customers' behaviors would change enormously, not only in the Share-riding sectors but also in other related industries, namely: retail, consumption, restaurant, entertainment. Customers’ information will be used to increase the capacity of AI and improve the customer experience and help third parties answer critical business questions, namely, where do the customers come from? or where should they run target marketing to acquire new customers? This information is valuable and could be hacked or sold to third parties. It raises the question about how customers’ data would be managed. Recently, New York City passed legislation that allows companies, namely: DoorDash, Uber Eats, and Grubhub, to share customers' information (name, phone number, email address, delivery addresses…) to restaurants. Despite the customers having the right to opt out of the program, this raises an alarming question about customers' privacy if they did not read through the opt-out question.

Truong Dinh