How Automated Trucks Could Create Better Truck Driving Jobs
Truckers keep America’s economy running. They bring the groceries to our local store, make sure the new smartphones arrive on time, and haul the fuel for our daily commute. Everything moves on a truck, and everything is moved by a trucker.
But truck driving jobs are getting harder. When you ask a driver about their challenges, you get a long list. On top of the day to day trials of the job — the long hours, the time away from home, the risk, the financial variability — there are plenty of new unknowns on the horizon.
One of the biggest questions is how automated trucks are going to impact truckers. In spite of the tough lifestyle truckers face, the industry provides many people with economic opportunity, freedom, and self-sufficiency. Will technology take that away? As attention to automation has grown in recent years, apocalyptic predictions have flourished: millions of truck drivers will be put out of work. Everyone from presidential candidates to the Onion is in on the story.
At Ike, we understand these concerns, and we’re working to build technology that can be complementary — not competitive — with truck drivers. There are important discussions to be had about automation generally, and its effects on our economy and society in the coming decades. But not all automation technology is created equal, and not every approach has to put people out of work. In fact, what’s unique about Ike’s model for automated trucks is that there’s an opportunity to create better truck driving jobs while saving lives and increasing the productivity of our transportation system.
We want to build technology for people. Over the last few months, we worked with economist Charles Hodgson to analyze Ike’s approach to automated trucking and build a detailed economic model that projects impacts to truck driving jobs nationally over the next ten years. Dr. Hodgson’s analysis builds on previous research to help drive a detailed and nuanced conversation about this important issue. We are open sourcing the entire model and all the data on GitHub, so others can build on top of this analysis and come to their own conclusions.
In summary: automated trucks can help shift difficult long haul jobs to short haul roles, keeping drivers closer to home and making use of their skills and expertise where it matters. Hodgson’s analysis suggests Ike’s approach could create nearly 140,000 new local truck driving jobs by 2030.
Read more below about the analysis or explore our interactive story map that summarizes the results.
Our approach: automating highway truck driving
First, it’s important to understand what we’re building at Ike, and what we’re not. Trucks powered by Ike’s technology will focus on highway driving, operating only in the simplest and most structured environment. As intriguing as it might seem to automate the entire journey of a truck (and driver), that’s just not realistic any time in the near future. Ride along on a trucker’s journey today and you’ll quickly see that truckers do a lot more than sit in the truck and turn the steering wheel. Especially at the beginning and end of the haul, drivers spend a lot of time out of the truck, using their hands, inspecting the vehicle, and working with other people to ensure everything gets to its destination smoothly, safely, and on time.
It’s not simple work. That’s why we’re focused on building driverless automated trucking for the highway. Trucks equipped with Ike’s technology won’t be able to back up to a dock, open trailer doors, talk to people, or even turn right at a stop sign, at least not for a long time. Instead, truck drivers will continue to do that complex and high value work.
In the future, when a trailer full of sneakers needs to get across the country, a trucker will move it to the highway and hand the load off to a driverless automated truck for the long haul. Then, on the other end of the journey on the highway, another driver in a regular truck will take the load to its final destination into town. That’s a great match between human skills and new technology that can also help make the industry safer and more productive.
This isn’t just the most practical technical approach, it also means truckers can have the opportunity to shift their work from the trying long haul routes to a “home daily” profile where they get to sleep in their own beds at night, work a more regular job, use their skills, and bear less of the risk and cost of the typical over the road driving job today.
By the numbers: quantifying the impact of automation
But how does this play out at large scale, when thousands of automated trucks are driving on the highways in the coming years? To help inform a more tangible discussion about Ike’s approach, we’ve worked with economist Dr. Charles Hodgson to build a statistical model of the impacts on truck driving jobs.
We have a great start on the analysis based on our leadership team’s prior work at Uber ATG. In 2018, they released an open source model to project the impacts of the transfer hub approach to automated trucking. Since then we’ve learned more about the industry and realized we could make some big improvements to the analysis.
First, previous analyses don’t account for how the varying distance of hauls impact the change in jobs. Drivers handing off to automated trucks for a 200 mile haul has a different effect than on a 2,000 mile haul.
Second, other work has overestimated the potential for automated trucks, assuming 1 million vehicles on the road in the next ten years, which isn’t realistic.
Third, we wanted to account for the fact that automation will roll out gradually in different geographies, not just suddenly be available everywhere.
Other interesting research has been completed in this area over the last few years as well, including a better estimate of the number of truck driving jobs by Maury Gittleman and Kristen Monaco, and support for the idea of “partial” industrial automation by Erik Brynjolfsson, Tom Mitchell, and Daniel Rock. We’ve also incorporated great feedback from our partners in the industry, who highlighted a few more key gaps in the analysis. So we asked Dr. Hodgson to build on top of the current research and rethink the implications of automated trucks.
The model works like this:
Understand the supply and demand for truck drivers
Using decades of historical data from a number of sources, Hodgson built a baseline picture of the economics of trucking. When the cost of something changes, it affects demand. When something gets cheaper, people buy more of it. When it gets more expensive, people buy less. But how much more or less? This varies depending on the good or service, and is known as elasticity.
Hogdson’s analysis concludes that the elasticity of demand for trucking services is 3.7, meaning that for every 1% cheaper freight becomes, 3.7% more will be purchased. The industry is quite sensitive to changes in price (economists say it is “elastic”). We can use this and other results to project how automated trucks might change the supply and demand balance in the industry. Hodgson also estimates the number of truckers that will retire in the coming years based on demographics and historical trends.
Make assumptions about how automated trucks will fit into the industry
For this model, we asked Hodgson to assume that automated trucks launch in Texas in 2023, but can only drive in certain conditions and at nearly the same cost as today’s operations. We assumed that the geography of deployment grows over the next several years, and that the performance of automated trucks improves until they are capable of navigating the entire interstate highway system in 2030.
Calculate how things may change in the future
By applying these assumptions to the baseline model, Hogdson was able to determine how many automated trucks would be in service at each point in time, and how many miles automated trucks would drive. Those miles replace long haul work that truck drivers would otherwise do. However, because trucks powered by Ike will need partners to move loads to and from the highway, all those automated miles also create local miles driven by truckers. Hodgson assumed an average of 25 miles of local driving on each end of a journey.
Depending on the distance of the haul, this has different impacts on truck driving work. For example, we asked Hodgson to assume that hauls under 100 miles are never automated (because the transaction costs are likely too high relative to just having drivers move those goods manually). On much longer hauls, the automated truck is driving most of the miles, so the translation to short haul miles is lower than in more regional applications.
Translate miles to jobs
The most important thing to understand about Ike’s approach is that automating highway driving does not mean truck driving jobs go away. Instead we see a shift from long haul to short haul. That kind of work isn’t for everyone, and there will be opportunities for drivers who love the open road for years to come. But for drivers who prefer sleeping in their own beds at night, relying on an automated truck to cross the country means a lot more local miles, and more time doing the things people do best.
What do the actual numbers tell us? Dr. Hodgson estimated how many individual jobs shift from long haul to short haul in this automated future by dividing the mileage projections described above by the number of miles truckers drive annually. For typical over the road jobs, truckers drive about 89,000 miles per year on average. For local jobs that number is more like 35,000 because of the slower speeds, traffic, and time at loading docks. By 2030, an automated truck could drive as much as 300,000 miles per year.
The results suggest something really interesting: while nearly 210,000 over the road jobs could be replaced by automation by 2030, 136,000 new short haul jobs would be created in their place — jobs that keep drivers closer to home and make better use of their skills. The net 73,000 job losses are more than offset by expected retirements over the next ten years (Hodgson estimates that 78,000 drivers will age out of the industry by 2030).
Feedback requested
As we’ve said previously, building automated trucking technology is hard, and there is a long road ahead. It’s important to be clear that this analysis isn’t a guarantee about the future. There are many assumptions embedded in this economic model, some of which may turn out to be wrong or need further improvement. We don’t have a crystal ball, and Ike will only be a small part of the future of transportation. But we hope providing these detailed numbers will make for a more nuanced discussion about the implications of automation throughout the logistics industry.
That’s part of the reason we’re open sourcing all of the code and the data from Dr. Hodgson’s analysis on Github. We welcome anyone to dig into the model, make their own assumptions, and see how their results may differ. This will be a useful toolkit for Ike, and hopefully others, to understand how one approach to automation may impact our economy in the future.
We’ve also heard loud and clear as we’ve consulted experts across the trucking world that technology alone is neither the problem nor the solution. There are all sorts of other issues that need to evolve alongside automation technology: industry structure, regulations, driver demographics, and more.
The big question we get asked after talking through this analysis is this: Will all that new short haul work be good jobs or bad jobs? Will they pay a living wage? We understand that concern, and we’re committed to doing our part to ensure that being a truck driver in the future is a safe, secure, valuable job. We’ll continue to work with our many partners in the industry to share our perspective and listen to concerns about how automation technology is used.
Trucking technology for people
One of Ike’s core principles is that technology is supposed to help people. We’re proud of our research collaboration with Dr. Hodgson, but to help make it more personal, we’ve spent time over the last few months talking to real truck drivers about their work, their challenges, and what the future may hold.
Anyone who works in logistics knows that there is no such thing as one kind of truck driving job. Nearly every role is different — where you drive, what you haul, who you work for, what kind of truck you use. The same is true for the impacts of automation. Some truckers will see new opportunities in the coming decade, while others won’t see any changes at all.
Here are five examples of drivers we have gotten to know, with our thoughts on how their jobs may change in the future. Each has a very different kind of job today, and very different implications for their work down the road. Click on a profile below to hear more from each driver:
Not all automation is created equal
The “future of work” is a hot topic these days, and for good reason. There are all sorts of difficult questions about what automation will mean for society, especially for people who work in blue collar jobs. At Ike we believe that the future depends on how technology is developed and used. We want trucking to be a case study in how to transform an industry in ways that benefit everyone.
We’ll have much more to share about our plans in the years to come. In the meantime, we hope this analysis will help advance the discussion about the impact of automation on truck drivers. We can’t afford to lose them, now or in the future.
