Are We There Yet?

-- By NathalieNoura - 17 Jan 2022

Optimism v Reality

In the past decade, companies have promised fully autonomous cars on the road by 2021 and that they would “almost end” car ownership by 2025. The US Secretary of Transportation promised the same in 2016. They predicted that by that date we would be able to call our car to pick us up, sit in a reclining seat sleeping or watching TV, and get dropped off at our destination, without traffic, accidents, or even a driver’s license. The proposition of highly automated vehicles is not unrealistic, but the expectations and timeline are. We are now in 2022 and not much has changed in our everyday driving, save for a few research autonomous cars. That optimism allowed companies to lure in talent and over 80 billion dollars in investments. Thus, the self-driving car became the most hyped technology experiment of the century. At present, the promise of self-driving cars (SCs) carries many concerns namely safety and privacy.

Only in the Suburbs

In 2019, Musk said again that Tesla autonomous robotic cars would hit the market by 2020. However, in 2021 he tweeted that “much of the AI in the real world has to be solved in order to enable unsupervised, generalized, fully autonomous driving”. This means that SCs are unable yet to drive in the real world in unanticipated circumstances, bad road infrastructure, numerous pedestrians, and extreme weather conditions. SCs rely on AI software, highly detailed maps, and high-tech cameras and sensors. AI software can recognize lanes and teach itself to stay within lines on a freeway. As such, it would have a hard time defining the lane when these lines are not visible, which is the case on many roads that are not periodically maintained. AIs are also unable to achieve a knowledge-based representation of the world. Even if engineers run detailed maps to fill gaps in sensor data, these maps are not updated enough to account for every possible scenario such as a new construction site which was not mapped. Would AI recognize a stop-sign if it is covered by a tree branch? Furthermore, visibility by sensors decreases in the rain, snow, fog, and severe weather conditions, which diminishes the effectiveness of those technologies. In October of 2021, Waymo started the very first autonomous taxi service in suburbs of Phoenix, Arizona. In the suburbs of Phoenix, the streets are wide, there are few pedestrians, and it mostly does not rain or snow, which makes it suitable for SCs. However, SCs are unable to function safely in a more complicated environment like in New York City which is crowded with pedestrians, bikes, and electric scooters, has tons of abrupt construction sites, and has downpours of rain and snow.

Autonomous or Controlled?

Another challenge presented by SCs is privacy. Cars rely on high-tech cameras and ultra-precise GPS data. As such, SCs collect large amount of data about operators, passengers, pedestrians, and other people recorded by the cars' sensors. The data includes camera images, activities, video clips, biometric data, user contacts, map data, location, speed, date and time, owner or passenger information, navigation history, audio and video content inside the cabin and its surroundings. For example, an abundance of location data provides a detailed record of the movements of the operator, passenger, or pedestrians recorded by the sensors that reveal features about their familial, political, professional, religious, and sexual associations like trips to the psychiatrist, plastic surgeon, abortion clinic, strip club, hotel, church, mosque or synagogue, gay bar, etc. Sensors also record private conversations inside the vehicle. Such private data could be accessed by hackers, private companies, and governments. Passengers can opt out of this invasion of privacy by refusing to get in the car, but phone conversations may still be recorded. However, pedestrians or people in other cars are unable to opt out. Such invasion of privacy, data collection, and surveillance affects individuals' autonomy and freedom. Furthermore, reasonable expectation of privacy applicable under the Fourth Amendment is diminished in SCs. Imagine if a cop can remotely turn off the car or retrieve the SCs data without a search warrant, based only on probable cause. Another risk would be to qualify operators of Levels 4 and 5 SCs as passengers rather than drivers and thereby strip them from the right to challenge a search and seizure warrant. Finally, all the data collected by private companies can be used to alter our behavior to maximize profits much like surveillance capitalism. In the futuristic world envisioned by companies, the SCs would suggest service providers, restaurants, and choose routes based on advertisements.

Did We Miss Something?

Beneath all the hype surrounding SCs promoted by private companies, something important gets buried in the background: autonomous busses, shuttles, and trucks. In 2021, Michigan City and Arlington, Texas had launched pilot programs for autonomous shuttles. France approved itsfirst shuttle to drive on public roads. Shuttles and busses can operate safely under controlled conditions: low -medium speed, maintained roads, minimal pedestrian traffic, and fair weather. The track, distance, speed and surroundings are the same, so the probability of unexpected circumstances decreases. Many other countries including Norway and China are also experimenting with fully automated public transportation of shuttles and busses. Privacy concerns with autonomous shuttles are mitigated. Autonomous shuttles are not tailored to individuals. They have a pre-automated route and do not present a challenge to the Fourth Amendment, any more than a regular bus would. It does of course require heightened attention to cybersecurity. Legislation can also be drafted to prohibit saving data collected by the shuttle or buss on pedestrians, and other third parties. Autonomous shuttles, busses, and trucks serve the exact purpose of SCs of decreased traffic, accidents, and pollution. It also provides added social benefits. More importantly, automated public transportation is feasible in the next few years, especially if capital is invested to that end. However, to reach a world where a person could fall asleep in his SC, which was promised by private companies driven by profits, could remain a dream for another decade. We are not there yet.