By Bill Gates, GatesNotes

I’ve always been a car guy. When I was younger, I used to love driving fast (sometimes too fast). Now, I look forward to my daily commute to work. There’s something so fun yet meditative about driving a car.

Despite that, I’m excited for the day I get to hand over control of my car to a machine.

That day is coming sooner rather than later. We’ve made tremendous progress on autonomous vehicles, or AVs, in recent years, and I believe we’ll reach a tipping point within the next decade. When it happens, AVs will change transportation as dramatically as the PC changed office work. A lot of this development has been enabled by the progress made in artificial intelligence more broadly.

Some background for those who might not know a lot about AVs: The best way to understand where we are today is by looking at the Society of American Engineers, or SAE, classification system. This is widely used to describe how autonomous a vehicle is.

In levels 0-2, a human driver is in full control of the car, but the vehicle can provide assistance through features like adaptive cruise control and lane centering. Level 3 is when the technology starts to move from the driver being in control to the vehicle being in control. By the time you reach the highest level, the car can be fully autonomous at all times and under all conditions—the level 5 vehicles of the future might not have steering wheels at all.

Right now, we’re close to the tipping point—between levels 2 and 3—when cars are becoming available that allow the driver to take their hands off the wheel and let the system drive in certain circumstances. The first level 3 car was recently approved for use in the United States, although only in very specific conditions: Autonomous mode is permitted if you’re going under 40 mph on a highway in Nevada on a sunny day.

Over the next decade, we’ll start to see more vehicles crossing this threshold. AVs are rapidly reaching the point where almost all of the technology required has been invented. Now, the focus is on refining algorithms and perfecting the engineering. There have been huge advances in recent years—especially in sensors, which scan the surrounding environment and tell the vehicle about things it needs to react to, like pedestrians crossing the street or another driver who swerves into your lane.

There are a lot of different approaches to AVs in development. Many vehicle manufacturers—like GM, Honda, and Tesla—are working on models that look like regular cars but have autonomous features. Then there are companies entirely focused on AVs, some of whose products are pushing the boundaries of what a vehicle can be—like a perfectly symmetrical robotaxi or public transit pods. Many others are developing components that can be installed to give an existing vehicle autonomous capabilities.

I recently had the opportunity to test drive—or test ride, I guess—a vehicle made by the British company Wayve, which has a fairly novel approach. While a lot of AVs can only navigate on streets that have been loaded into their system, the Wayve vehicle operates more like a person. It can drive anywhere a human can drive.

When you get behind the wheel of a car, you rely on the knowledge you’ve accumulated from every other drive you’ve ever taken. That’s why you know what to do at a stop sign, even if you’ve never seen that particular sign on that specific road before. Wayve uses deep learning techniques to do the same thing. The algorithm learns by example. It applies lessons acquired from lots of real world driving and simulations to interpret its surroundings and respond in real time.

The result was a memorable ride. The car drove us around downtown London, which is one of the most challenging driving environments imaginable, and it was a bit surreal to be in the car as it dodged all the traffic. (Since the car is still in development, we had a safety driver in the car just in case, and she assumed control several times.)

It’s not clear yet which approaches will be the most successful, since we’re only starting to reach the threshold where cars become truly autonomous. But once we get there, what will the transition to AVs actually look like?

For one thing, passenger cars will likely be one of the last vehicle types to see widespread autonomous adoption. Long-haul trucking will probably be the first sector, followed by deliveries. When you finally do step into an AV, it will likely be a taxi or a rental car. (Rental car companies lose a lot of money every year to driver-caused accidents, so they’re eager to transition to an AV fleet that is—at least in theory—less accident-prone.)

As AVs become more common, we’re going to have to rethink many of the systems we’ve created to support driving. Car insurance is a great example. Who is responsible when an autonomous vehicle gets in an accident, the person riding in the car or the company that programmed the software? Governments will have to create new laws and regulations. Roads might even have to change. A lot of highways have high-occupancy lanes to encourage carpooling—will we one day have “autonomous vehicles only” lanes? Will AVs eventually become so popular that you have to use the “human drivers only” lane if you want to be behind the wheel?

That type of shift is likely decades away, if it happens at all. Even once the technology is perfected, people might not feel comfortable riding in a car without a steering wheel at first. But I believe the benefits will convince them. AVs will eventually become cheaper than regular vehicles. And if you commute by car like me, just think about how much time you waste driving. You could instead catch up on emails, or read a good book, or watch the new episode of your favorite show—all things that are possible in fully autonomous vehicles. More importantly, AVs will help create more equity for the elderly and people with disabilities by providing them with more transportation options. And they’ll even help us avoid a climate disaster, since the majority in development are also electric vehicles.

Humanity has adapted to new modes of transportation before. I believe we will do it again. For most of our existence, we relied on natural ways of getting around: We walked, or rode on horseback, or traveled in a boat pushed by wind. Then, in the 1700s, we entered the locomotion age when mobility was powered by steam engines and internal combustion. Now, we find ourselves in the early days of the autonomous age. It’s an exciting time, and I can’t wait to see what new possibilities it unlocks.

By Beth Stackpole, MIT

It’s a familiar narrative: Robots and other forms of automation, while good for driving business efficiencies, are a death knell for the modern workforce.

Yet introducing robots to the workforce could serve as a positive force, delivering benefits to corporations as well as to the workforce and global economy.

“The zero-sum [take] is that companies become more profitable, more productive, even more flexible, but the consequence is that workers get displaced and are less valued,” said Ben Armstrong, a research scientist and executive director at MIT’s Industrial Performance Center, at the 2022 MIT Digital Technology and Strategy Conference.

“We don’t see automation displacing workers — in fact, it’s the opposite,” he said. “Firms that invest in automation equipment like robots or advanced software end up being more profitable and hiring more people.”

There has actually been too little automation, Armstrong contended. Slightly less than 10% of U.S. manufacturers have deployed industrial robots, due to design limitations, workforce challenges, and high integration costs. Slow productivity growth among small and midsize companies over the past few years is one indicator that there might not be enough technology deployed to drive competitive advantage, he said.

In a 2021 report, the International Federation of Robotics indicated that installations of new robots grew only slightly in 2020 worldwide, except in China, which saw a 20% increase. The latest report, however, shows growth in robot installations in all major markets and regions, including a 31% increase in North America and a 24% increase in Europe in 2021.

Achieving “positive-sum automation” starts with understanding why it can be so challenging for companies, as well as creating more flexible tools, Armstrong said.

Barriers to robot adoption

Robot adoption has remained flat for several reasons, said Armstrong, who co-leads the MIT Work of the Future initiative. The biggest inhibitor by far is a lack of digital skills, which are essential for building a workforce that’s able to effectively configure, operate, and repair robots and automation systems. Proficiency in specific programming languages and other skills remain scant in manufacturing companies of all sizes. Only 18.6% of new production-related jobs in manufacturing have a digital skills requirement, Armstrong said, which means few incoming workers are qualified to optimize robots and automation for true business advantage.

Companies are also struggling with robot flexibility. Organizations need to be able to continuously adapt robots and automated systems to changes in the environment. Most robots and automation tools function as black box systems, which makes it next to impossible for the average worker to understand how the tools operate, let alone seamlessly switch between systems to get a specific task done.

“We see positive-sum automation as overcoming these barriers with technology applications that deliver productivity gains,” Armstrong said. “The goal is for a high-throughput, highly repetitive automated system that can also improve flexibility, has an ability to switch between products, and can add innovations and adapt as the business grows and changes.”

A path to productive automation

Armstrong and his colleague Julie Shah, an MIT professor who leads the Interactive Robotics Group of the MIT Computer Science and Artificial Intelligence Laboratory, have developed a three-pronged, multidisciplinary approach aimed at reducing complexity and helping organizations achieve more widespread and productive automation. Their approach addresses the following elements:

Design. Most robots are difficult to program and especially hard to reprogram to worker specifications, which means they often remain unused. Armstrong recounted the struggle of a midsize Ohio-based manufacturer that invested in a collaborative robot so it could add another shift without the pain of finding workers in a tight labor market. After consultants completed the initial configuration and setup, there was no one on staff who could program the robot, which left it sitting idle. Large companies like Tesla have also had well-documented struggles operationalizing robots and implementing so-called lights-out manufacturing. There are reports of robots breaking delicate parts as well as causing other miscues that require human intervention to get operations back on track.

Armstrong said companies should look for robot designs that are more accessible and adaptable. The addition of an intuitive user interface and low-code capabilities will allow workers to adapt robots to the external environment and make changes on their own without having to hire expensive consultants or incurring high switching costs. In this scenario, robots can work in tandem with technicians to assemble equipment — with the robot manipulating heavy assembly while the human performs the dexterous work, for example. The upside is more productive assembly that improves overall job quality.

Integration. Historically, robots and automation system have required specific programming skills and a reliance on external partners for programming and maintenance throughout the life cycle. Taking a bottom-up deployment approach can shift this dynamic and empower internal workers to step in as the experts. Armstrong cited a hospital system that employed robotic process automation software to automate mundane administrative tasks. Instead of having C-level executives decide what processes to automate, the hospital took a bottom-up approach, allowing internal personnel responsible for the tasks to have input and receive training to program the software themselves. The result was administrative teams who were thrilled to avoid the “soul-sucking” work of routine tasks and instead focus on the work they liked most, Armstrong said. Executives, having achieved more efficient operations, were happy as well, he added.

Measurement. To get around the black box system limitations, it’s important to combine investment in automation technology with a comparable investment in employee training. One small manufacturer trained workers and then linked overall machining hours to bonuses. The more streamlined the processes were, the more likely it was that the employee would receive a bonus. This change, coupled with a wage hike to compensate for additional skill requirements, resulted in more productive automation.

Armstrong said that by embracing the concept of flexible automation, organizations will get more mileage from their investments. In scenarios with repetitive, high-volume tasks or labor shortages, robots can serve as a more productive workforce. But automation shines when robots and humans work together as a team to deliver high-quality products to customers or innovate new products and processes.

“With more flexibility, robots and software become tools that help teams become better at tasks,” Armstrong said. “They aren’t just doing the highly predictable tasks — they are also adjusting as the team figures out better ways to solve problems.”

By David Taube, Transport Dive

Former USA Truck CEO James Reed recently shifted into the role of COO at autonomous truck tech company Kodiak Robotics.

Reed talked with Transport Dive about his background, Kodiak’s autonomous trucking technology and his decision to join the AV tech firm.

The executive talked about how autonomous trucking could improve lifestyles for drivers, delved into the differences between AV tech approaches and explained why he placed his “career bet” on Kodiak.

TRANSPORT DIVE: Why were you drawn to Kodiak?

JAMES REED: Our company was exploring the autonomous space, and we weren’t working exclusively with Kodiak. So we were moving freight with some autonomous vehicle providers, just trying to dip our toe in the water and understand it, and we had formed a really good relationship with Kodiak.

On the merits of ability to go to market, I thought Kodiak’s mirror pod solution, as far as mapping goes, was frankly the most scalable and maybe even the most clever way to go to market. So I was like, ‘If I’m gonna place a bet, this is where I’ve placed my career bet.’ And I really liked the team, too.

TRANSPORT DIVE: How you are you seeing Kodiak stack up against its competition? Could you elaborate on Kodiak’s technology?

REED: It’s been Kodiak’s intention from the get-go to move sensor suites into the mirror pods. Being able to replace a pod quickly and efficiently and keep the truck moving is a lot easier than the alternatives.

Many in the industry are using high-density mapping. Once they create that fully featured map, the vehicle essentially drives, which is great — until the map’s wrong.

The technology needs to recognize two things. First, it needs to recognize that the map doesn’t match what it’s ‘seeing’ — I’ll use that word. And then it needs to perceive what it’s seeing and devise a reaction to it. So at some level, perception will be required of most solutions. That’s my layperson understanding of it.

At Kodiak, we define lane parameters. So we’ll drive a lane. In our mapping, we will map the lane markers. And then we deploy a vehicle shortly thereafter. And that vehicle can go in autonomous mode, understanding where the lane markers are, calculating a forward path every millisecond to predict where it should go. So it’s really a low overhead way to deploy perception into the autonomous space. It’s pretty brilliant.

TRANSPORT DIVE: There’s been this tension between new technology and what that might do to jobs. Have you seen this resistance play out firsthand?

REED: One of the interesting innovations in trucks was when when automatic transmissions came into long-haul trucking. So fast forward into the early 2010s: It’s really hard to recruit a driver. Some of the newer drivers didn’t know how to drive anything but an automatic. … And for fleets, it was cheaper to run automatic transmissions, better fuel economy, you could get more drivers because you get these guys and gals with automatic designations on their CDLs into the trucks. And the old timers hated it, saying ‘I’ll never drive an automatic.’ Well, I can give you names of dozens of drivers that have been driving 30–40 years, who now wouldn’t drive anything other than an automatic.

I’m not saying that’s a perfect corollary for autonomy, but today, we have a driver shortage. It’s approaching 80,000 drivers and projected to worsen. As you look forward, we’ve got an opportunity in the transportation industry to ameliorate and to alleviate the challenge around driver recruitment by supplementing it with autonomous vehicles. You’re much more likely to push jobs to the types of jobs that drivers like, such as being home daily. Local jobs are going to be much more prevalent, especially if we can fill that middle mile section, which is our intent with autonomous technology.

TRANSPORT DIVE: If other executives are considering a similar move from traditional trucking companies to self-driving companies, what advice do you have for them?

REED: It’s the same advice that I have for investors. It’s the same advice that I have for potential customers: Get educated and form an opinion. I mean, my choice to join Kodiak was a well-informed choice that’s taken five years to develop an opinion about the industry. I would caution them to be careful to make sure that this is a space that they understand and that they want to have a voice in. I truly believe in the technology. I believe it’s further along than most people realize.

If somebody thinks they’re wired that way, by all means, call Kodiak first.