Self-Driving Cars Have A Long Road Ahead (Part 3)
Our destination takes us deeper into the entangled considerations for our society, environment and ethics.
In part one of this series, we explored the role of human error in road accidents, with over one and a half million crashes in the US annually. It highlighted how self-driving cars stand to reduce these incidents, alongside the economic and social challenges they present, such as job displacement and retraining needs.
In part two, the evolution of Autonomous Vehicles (AVs) was explored. The previous letter highlighted advancements spearheaded by Tesla and Google, as well as illustrating the progression and challenges of the technology, such as the different levels of automation in cars.
This third and final piece covers a more forward-thinking approach, epitomizing the spirit of Hyperopia. All the technological advancements will be weighed against societal concerns, with a focus on geopolitical and environmental impact. New mediums of technology reshape the environment and society they are used in. This means the impact of AVs will be seismic. To begin, this letter sets out a few thought experiments to expose ethical dilemmas. The purpose of these thought experiments is to help unveil some key challenges, that regulators and lawmakers must face in these complex situations.
States across the U.S. have been actively passing legislation related to AVs. The laws vary, with some states allowing testing without a safety driver and others focusing on commercial applications like truck platooning. The differences in state laws, such as the definition of “vehicle operator,” have implications for how AVs are licensed and regulated. This patchwork of regulations across states reflects the ongoing experimentation and adaptation as AV technology continues to evolve. For better or for worse, this has been shaped largely by Big Tech.
Tech companies work closely with lobbyists. Google, through Waymo, has played a significant role in shaping AV legislation. Different countries may have competing interests in shaping these standards, reflecting broader geopolitical tensions. It may not be so much of an issue of safety or sustainability, but lobbying that brings AV to us. I’d rather not have profit-orientated, hyper capitalist Big Techtroglodytes decide laws that have such a drastic on daily life. Instead, these laws should be influenced by educated citizens, putting pressure on politicians and fighting for representation in public forums and lawmaking. On that note, a series of thought experiments may be a good way to illustrate challenging legal questions.
Imagine a narrow tunnel. A child suddenly dashes towards the entrance. With no room to swerve, the driver must make a grim choice: to risk the child’s life by continuing forward, or to crash into the wall, jeopardizing the life of the occupants. Now, imagine a car in a tight spot on a highway, say, trailing behind a motorcyclist. A sudden crash ahead sparks a chain reaction, forcing the car into a dire choice. Finally, just in case either scenario seems unlikely, imagine a pedestrian crossing the street during a red light. Should the car stop, risking a rear-end collision, or swerve and risk the occupants’ lives? All of these are variations of the Programming Priority Dilemma.
These philosophical questions, perhaps unanswerable, will need to be addressed by the designers of AVs. That leads us to another dilemma: the responsibility and accountability of the accident. Who is responsible? The owner of the car? The software engineer? The road and safety administration? For those of you into philosophy, I’ll throw in another thought experiment: imagine what will happen when your car could be hacked, controlled and used for malicious intent against your wishes.
While you may consider these to be hypothetical examples, there is a compelling case to be made that the physical world is about to be reshaped to fit the new medium of AVs. Consider suburban expansionism after World War II. Many farmlands were transformed to bustling suburbs. From 1940 to 1970, suburban populations in the US rose over 250%. This outward sprawl was enabled by the flexibility and convenience cars offered over public transport. There will likely be a push towards more pedestrian-friendly areas, due to a reduction in lanes needed and the overall efficiency of the transport systems. The benefits are innumerable, but often, changes in infrastructure are driven by political agendas. Governments, when given the chance to build over roads and park places, tend to push out low-income residents and small businesses out, in the name of gentrification.
Taking New York as an example, there is a silver lining to a successful implementation of self-driving cars in a large-scale city. Research from the MIT SENSEable City Laboratory suggests that by combining ride-sharing with car-sharing, cities like New York could potentially meet passenger transport needs with 80% fewer cars and free up space equivalent to about 900 city blocks currently used for parking by encouraging the use of robo-shuttles. This would significantly reduce congestion and the space required for parking, allowing for more diverse urban land use. Not to mention the reduction in carbon dioxide emissions. Los Angeles stands to reduce its carbon dioxide emissions by millions of metric tons annually by promoting shared AVs and curbing its private vehicle fleet.
Cars are changing to become multipurpose spaces on wheels, possibly even extending to mobile clinics or other services. This shift reflects a broader trend towards reimagining how vehicles are designed and used in line with emerging technologies and societal needs.
Looking away from the US and any one specific nation, the growing market for AVs stands poised to instigate impact in every corner of the world. By making longer commutes more manageable and reducing the stress associated with driving, AVs could encourage a further expansion of suburban living. This potential for a new wave of suburban growth echoes the post-WWII era, suggesting a reshaping of the American dream in the 21st century, focused once again on suburban life but driven by cutting-edge technology. On the verge of possibility is the dramatic alteration of global trade.
When maritime routes were shortened between Europe and Asia, because of the Suez Canal’s opening, seaborn trade nearly doubled. Its impact ricocheted outwards, spurring colonization and influencing everything from ship design to commerce to empire building. AVs have a similar potential to revolutionize land-based trade and logistics. By optimizing supply chains and introducing new efficiencies in transport, they could open up new trade corridors, reminiscent of how the Suez Canal created new maritime routes. Such trade routes could shake up regional economies and global trade dynamics. Or, take the simple elevator, which paved the way for modern skyscrapers. AVs will challenge our perception of distance and travel. This shift could lead to more dispersed urban sprawls or new urban designs prioritizing AV mobility, much like cities were reimagined to accommodate skyscrapers.
The introduction of AVs may not only change how we travel but also how we live, work, and interact within our urban environments. The sheer volume of data will create many jobs, driven by collaborations between governments, companies, and universities. Such data promises to boost the utility of AI models, incentivizing the deployment of more AVs.
A single AV can generate approximately four terabytes of data per day, including visual information from cameras, LiDAR and radar data, GPS and mapping information, and sensor data for vehicle performance. Privacy then becomes paramount, as much of that data can be used to identify specific people.
The successful integration of AVs will require new physical and digital infrastructure, as well as urban planning. This includes dedicated lanes and sensors that enable self-driving cars to communicate with their surroundings. In some scenarios, promoting other mobility options like e-bikes and e-scooters could be more beneficial than introducing AVs. Cities might evolve to become more pedestrian-friendly, reducing the need for parking spaces. This change could lead to a redistribution of urban space, affecting real estate values and urban lifestyles. The transformation of cities like Amsterdam from car-centric to bike-friendly over the past decades provides a historical reference for such shifts.
This brings us environmental damage from the advent of AVs. They lead to high energy consumption and increased emissions, particularly when the electricity that powers them is sourced from fossil fuels. The U.S. Energy Information Administration has noted that the widespread adoption of AVs could potentially lead to higher transportation energy consumption. One study MIT researchers indicated that if one billion AVs each drove for one hour per day, the energy consumed by their onboard computers would equate to the current energy consumption of data centers globally. Another study from the University of Michigan found that the sensors and computers in a self-driving car could consume between two to four kilowatts of electricity during operation. This additional power draw could reduce the overall efficiency of an electric vehicle by 5% to 15%. Though it is worth considering how lower costs over time could offset this. After all, it is unlikely that the hardware will not improve; and software can become more mobile and resource friendly- although the rush towards ubiquitous AI will make us tend towards whatever is available in the moment.
Training the AI models that power AVs is a highly energy-intensive process. Although in most cases, you only have to train it once and then it can be re-used, and deployed for serving user requests. Serving and deploying it is not free either. Such models drive demand for large-scale data centers, which are energy-intensive operations. In AI training, the computational power used has been doubling roughly every three and a half months since 2012, exacerbating the energy consumption problem. By 2030, computers are projected to consume between 8-21% of the global electricity supply, up from 1-2% in 2018. Furthermore, Hardware components like sensors and processors used in AVs have a finite life cycle, contributing to electronic waste when they become obsolete or fail. Urban sprawl is another concern as AVs could encourage living further from urban centers, potentially leading to increased land use for housing, longer trips, and consequently, higher energy consumption and emissions. At the same time, this could offer something like a reversal of the modern state, which plans urban layout to favour movement to and from major cities, especially capitals, over interregional or local traffic.
AVs could significantly disrupt traditional automotive industries, a cornerstone of the economy in countries like the United States (Detroit), Germany (Wolfsburg, Stuttgart), and Japan (Toyota City). This disruption could lead to job losses in manufacturing sectors, leading to social and economic repercussions. Jobs will also be hit particularly hard in commercial sectors like trucking. As AVs disrupt traditional jobs, new labor movements may emerge, demanding protections and reforms similar to the disruption of the automobile revolution.
In the United States, the Executive Order on Maintaining American Leadership in Artificial Intelligence, signed by President Trump in February 2019, lays out a broad strategy for AI development and regulation. While it primarily focuses on maintaining leadership in AI, it also touches on issues such as AI governance, safety, and fairness, which can indirectly influence AI's sustainability and carbon footprint. Since then, the Biden Administration has proposed an Executive Order. When this was announced I began to cover this, with my latest piece available here.
The state of California has introduced regulations on the energy efficiency of electronic devices, including Graphics Processing Units (GPUs) used in AI. Canada's Artificial Intelligence and Data Act, a component of Bill C-27, updates the country's privacy laws and reflects a growing consciousness of AI's implications. This legislation, along with other measures, targets AI and data regulation, emphasizing privacy and sustainability.
The European Union's Artificial Intelligence Act, adopted in May 2023, represents a significant growth in AI regulation It targets high-risk applications, emphasizing consumer protection and categorizing AI systems into risk tiers. This may avoid the training of large scale models, for example, something like GPT-5. Training such a large model has a marked environmental impact. While it addresses safety and ethics, the Act is less focused on environmental sustainability, forming part of the EU's broader legal framework for AI. Italy has been active in regulating AI, especially in the context of the European Union's broader strategies. The country aligns with the EU's regulations, including the Artificial Intelligence Act, focusing on ethical and sustainable AI development. In the United Kingdom, the AI Strategy released by the Secretary of State for Science, Innovation, and Technology aims to establish the country as an AI leader. Then there are data center regulations in places like Frankfurt, Germany, and Groton, Connecticut, USA, address the energy demands and environmental impacts of these key AI infrastructure components. These measures indirectly influence AI sustainability by governing the energy use and carbon emissions of data centers involved in AI training.
In Asia, China's Administrative Measures for Generative Artificial Intelligence Services, there is a focus on alignment societal morals and avoidance of discrimination. This regulation underscores accuracy and intellectual property respect in AI outputs, indirectly aiding sustainable AI practices.
The advent of automobiles significantly increased global demand for oil. This turned countries in the Middle East, rich in oil reserves, into geopolitical hotspots. Their oil wealth attracted exploration and investment from Western countries, particularly the United States and European nations. With China now a key player in modern geopolitics, their rivalry with the US in AV technology will be a tumultuous one.
China's targeted investment in the EV and AI sectors as part of its "Made in China 2025" plan demonstrates a strategic shift to dominate future technology markets. This move is seen as a direct challenge to U.S. technological leadership, potentially reshaping global technological standards and supply chains. This divide echoes the Cold War, with two humongous nations vying for supremacy in AV technology, akin to the space race or nuclear arms race. In general the rise of electronics has escalated the demand for REMs (rare earth minerals).
China’s growing exports may force the US and those in the European Union to either develop new batteries or finding an alternative source of REMs. Countries with significant REM reserves, many of which are in regions with historical political instability or underdevelopment, including many countries in Africa and South America have become focal points of geopolitical interest. Not just for exploration and mining but also for securing supply chains and influence in these regions. Through China’s Belt and Road Initiative (BRI), the country aims to expand into lithium-rich countries, such as Bolivia, Argentina, and Chile. This expansion is not just economic but also a strategic move to secure vital resources for AV technology.
The integration of AI and AV technologies in military applications is a key facet of the U.S.-China technological rivalry. The development of autonomous drones, unmanned vehicles, and AI-integrated military operations by both nations illustrates the strategic importance of these technologies in future warfare scenarios, as well as the potential for a new type of arms race focused on autonomous and AI-enabled systems.
The US Army's Next Generation Combat Vehicle program includes an optionally manned fighting vehicle and a family of autonomous robotic combat vehicles. As AV technology gets better, it may abet military uses. It’s worth noting the sheer amount of money that could scale up these AI-based technologies if military investment continues. As with many military contracts, the government hire software and hardware suppliers through the Pentagon. In another piece I explore the role the military plays in the US economy .
Nations continue to develop cyber warfare capabilities targeted at disrupting AV systems, which becomes a new frontier in international espionage and conflict. The U.S. and China might establish military bases near key lithium reserves, evoking the strategic military placements seen during the Cold War. You can read more about other questionable applications of AI in military operations in the link below.
The conversation we have about AV technology and ethical mining of REMs could become a central topic, with representation at the G7/G20 Summits on AVs and REMs. In recent years. Global environmental movements, akin to the protests against oil pipeline projects like Dakota Access, or to increasingly military influence in the Middle East, could arise in response to the environmental impact of lithium mining.
This is something I will explore further in a future piece, focusing on the ecology and geopolitics of rare earth minerals and other resources used in modern information and AI systems. In the meantime, read on to learn more about human, AI and drone podcast.
Unlike many examples of computer-human pairings, humans with AVs may increase the risk of the AV making a mistake. At the same time, it’s not clear what the error rates for totally self-driving or even level 4 automation would be. When the conditions have stabilized, it may be human-computer hybrids prove to be the best choice of driver too.
Next, check out my piece on self-driving drones, delivering right to your doorstep.