Just as the automobile shaped much of the urban and social fabric of the modern era, autonomous vehicles promise to be a major determinant for how society is organized for decades to come. Because our decisions on how they are built have long-term, irreversible implications, it is essential to craft the right incentives behind development and operations.
1. Incentives shape technology
Incentives shape whether technology makes humans better off, worse off, or, in most cases, a mix of the two. Incentives—whether or not consciously considered—nudge the builders of technology on what to prioritize, who to build for, and how the experience should be shaped. This can cause one underlying technology to lead to dramatically different systems of use.
Incentives have shaped many systems in our modern world that we take for granted. For example, since the industrial age the owners of duplicators (machines producing low-cost, standardized products) were incentivized to maximize their utilization irrespective of the utility of their use, which led to the discipline of marketing—creating demand for an oversupply of goods. Similarly, the business models of contemporary digital platforms are often based on maximizing user engagement, leading to sophisticated tactics that manipulate innate human biases, causing over-engagement on their platforms. Interestingly in early 2018, Mark Zuckerberg announced that Facebook will prioritize designing for “time well spent.” How the company navigates the tension between maximizing engagement and “time well spent” remains to be seen.
2. The case of the automobile
Since autonomous vehicles (AVs) promise to be the predominant mode of next-generation mobility, establishing the proper incentives and guardrails is crucial. We should be cautious because the way in which AVs and their accompanying infrastructure are built will have effects that could last for decades. In general, once capitally-intensive systems are put in place, it becomes difficult to reverse or correct past decisions (e.g., the economics of plastics are based on previous investments in oil refineries). This suggests that municipalities, nations, and companies may adapt mobility infrastructure to changing priorities by making incremental rather than fundamental changes.
“… now there’s a six lane highway separating my leafy neighborhood from downtown Detroit. The i-375 highway may have seemed like a good at one point in time, but so did using a stylus to operate a cell phone. Replacing an outdated cell phone takes minutes, but doing the same for mobility infrastructure can take decades.” — Bryan Boyer
The story of the automobile in North America is a cautionary tale. The liberating nature of the automobile—freeing people from the limitations of their geography—led to a set of policies that compounded the incentive of automakers to maximize number of vehicles sold. The subsidization of highways and roads failed to capture the true costs of owning and operating an automobile, making it artificially cheap relative to alternative modes of transport (e.g., trains and streetcars). The economics of assembly line manufacturing incentivized every automaker to maximize their utilization— spurring aggressive marketing to boost demand. The result was a car-centric built environment that we continue to live with more than 50 years after the foundations were set.
The dream of an automobile—a 4,000 lbs object—for every adult did not work at scale. Within cities, despite many municipalities prioritizing policies to support cars rather than public transit, their infrastructure failed to keep up with the growing number of cars on the road. An irony emerged: high volumes of people becaming relegated to tight sidewalks, while spacious streets filled with low-occupancy vehicles became stuck in congestion. Outside of cities, subsidized roads and low transportation costs accelerated the outward expansion of populations into low-density suburbs. The car became a requirement for getting around, in turn causing more cars to go on the road.
Ironically the infrastructure of cars, built decades back, will serve as the tracks on which AVs are built, establishing a new infrastructure for decades forward.
3. The great race for AVs, and the race to respond
Companies are rushing to be the first to commercialize autonomous driving technologies—with each competing for a share of what Intel forecasts to be worth $800 billion in 2035, and $7 trillion in 2050. This race is characterized by “coopetition” along the value chain. The players include: vehicle manufacturers (e.g., Ford, GM, Tesla), providers of autonomous systems (e.g., Waymo, MobilEye, Tesla, TomTom), ridesharing and transportation network companies governing vehicle utilization (e.g., Uber, Lyft, Didi Chuxing), and the infrastructure on which vehicles sit (e.g., municipalities, states, private toll highways).
Each player has its own profitability levers (i.e., incentives), which are optimized to recoup capital costs and earn returns for investors. Here are a few of those levers:
- Vehicle Manufacturers: number of vehicles sold, average margin per vehicle
- Transportation Network Companies: number of trips completed, average km/mi traveled per trip, average margin per km/mi traveled
- Data Aggregators: mobility data mapped to online behavioural data, value of “insights” for the aggregator and its customers
While these levers are essential for realizing the benefits of AVs, it’s important to understand their potential effects and externalities on systems when maximized. To illustrate how different incentives surrounding AVs could result in different manifestations of the technology, I’ve outlined three brief, fictional scenarios—these are not intended to be predictive nor comprehensive:
A/ The Car Continues: Imagine that automakers stay in the driver’s seat by acquiring their own autonomous capabilities (“autonomy in a box“), and partnering with ridesharing companies to route and optimize their vehicles on the road—whether these roaming AVs are owned by wealthy individuals or accessed by the public. As manufacturer-operators, they maximize number of vehicles on the road and the utilization per vehicle. This future is most like the existing paradigm: with manufacturer-operators continuing to maximize production of their cars. While higher vehicle utilization could alleviate some congestion in urban areas, manufacturer-operators follow the same trap of the last century: to overuse and neglect their effect on roads—the underlying “free” public infrastructure.
B/ Mobility Fever: Imagine that ridesharing companies dominate the market by deploying their own AV fleets, with automakers relegated to fulfilling hardware orders. Ridesharers continue to tweak their shape and performance of their AV fleets to maximize the number of trips—whose dollar value depends on distance and up-sells to premium options. At scale, the costs of transportation fall and public transit withers back to its core of subways and streetcars. At best, this rideshare-centric future delivers a host of socioeconomic benefits from enabling endowing greater mobility. At worst, individualized modes of transport, like the car, further congest streets as transit usage drops. In addition, sprawl could further proliferate as the hurdles for living further from work drop—falling costs of transportation and heightened comfort within the AV.
C/ A Balancing Act: Imagine that municipalities formulate a mobility strategy that aims to curb congestion, support urban planning, and increase availability of mobility to all citizens. The municipalities levy a dynamic road usage tax that charges each AV based on distance traveled, street congestion, location, time, fuel type, vehicle dimensions, and vehicle occupant density. Irrespective of whether manufacturers or ridesharers are at the helm, they modify their fleet mix and algorithms to maximize profitability within the new constraints. For instance, manufacturing could shift towards fleets of mid-density AVs that carry 10 people at a time and consume less space than individualized transports. At best, externalities such as congestion are contained. At worst, incorrect or shortsighted assumptions made by policy makers could drive unintended consequences.
4. Shaping the new system of mobility
While there is a rush to develop the autonomous vehicle, there is still runway to proactively shape its development before it supersedes the automobile as the predominant mode of transportation. As the venture capitalist Benedict Evans usefully summarizes, expert forecasts for adoption of Level 5 (fully autonomous) AVs range from 5–10 years. Additionally, it will take time for non-autonomous vehicles to be phased out, and for AVs to reach geographically dispersed markets. This suggests that there will be a transient period where autonomous and non-autonomous vehicles coexist on the roads.
Part of the solution requires (i) understanding the incentives surrounding AVs, (ii) identifying systemic effects resulting from incentives, and (iii) formulating a response to disincentive negative externalities. Developers of AVs should critically review how their profitability levers relate to broader systems and feed these learnings into the design process. Policy makers should immerse themselves in the AV ecosystem and understand how differing incentives drive differing systemic effects. Experts across disciplines—such as architecture, urbanism, anthropology, corporate strategy, and foresight—should be engaged to fill the gap in understanding how AVs impact other domains.
The above is only a sketch of ideas that might be worthy of consideration. However, it alerts us to the necessity to understand and shape the incentives behind the development and commercialization of technology. By advancing such efforts early, the reification of autonomous vehicles can avoid undesirable futures that stem from a failure to anticipate and contain externalities.