A Blog by Jonathan Low


Jan 28, 2022

How AI Is Now Enabling Even Physical Jobs To Go Remote

Remote workers can operate multiple forklifts and other industrial equipment thousands of miles away by using AI-enabled technology.

These advances come as the US logistics industry alone is short over 500,000 workers. While there will continue to be safety concerns, this mode of operation seems destined to continue and may, eventually, through machine learning, render even the remote operators obsolete. JL 

Will Knight reports in Wired:

An operator of a forklift in France—while sitting behind a desk in California - moved a stack of pallets into a waiting truck using a setup resembling a video gaming rig. He sat behind a steering wheel and pedals that transmitted commands to the forklift thousands of miles away; large screens offered views in front, behind, and to the sides of the vehicle. New forms of remote physical work involve supervising multiple autonomous machines. (But) some companies see remote human labor as a temporary way to train AI algorithms that will eventually replace them.

ERIC MCCARTER REMEMBERS the first time he operated a forklift truck in France—while sitting behind a desk in California.

McCarter used the forklift to move a stack of pallets into a waiting truck using a setup resembling a video gaming rig. He sat behind a steering wheel and pedals that transmitted commands to the forklift thousands of miles away; large screens offered views in front, behind, and to the sides of the vehicle.

The vehicle relies on limited artificial intelligence to avoid obstacles and safely come to a stop if the connection between France and the US were to fail. But the AI isn’t yet clever enough to let the robotic forklift navigate on its own through an unfamiliar warehouse or take on a new task.

The work was surprisingly easy, McCarter says, although he found it a little tricky to keep track of the vehicle’s position relative to nearby obstacles. “It's definitely challenging to operate in a facility that you've never physically set foot in,” he says.

McCarter works for a company called Phantom Auto. He was testing the company’s remote technology that lets a driver operate a forklift without physically sitting in the vehicle. Two major logistics companies, ArcBest and NFI, plan to begin using the tech later this year, hoping to tap a wider pool of forklift drivers who live far from a specific warehouse. The warehouse and transportation industry had a record 597,000 job openings in November, according to the US Bureau of Labor Statistics.

Many office workers have gone remote during the pandemic, with some firms making the shift indefinite. But the continued challenges posed by the virus and a deepening labor shortage—combined with advances in technologies such as AI and virtual reality—are allowing a small but growing number of physical jobs to go remote as well.

“There's just no way this isn't going to explode as a category,” says Matt Beane, an assistant professor at UC Santa Barbara who studies human-robot workplace collaboration. As industrial machines become more capable and connected, Beane says, the number and variety of these jobs will grow.

Beane says new forms of remote physical work may involve supervising multiple autonomous machines, similar to how one person oversees several AI chatbots simultaneously. Chatbots can handle many common questions but will lose the thread when the conversation veers off topic or becomes complicated, annoying users.

The approach can help companies find new workers and create more efficient processes, but it could be less positive for workers. Some remote jobs are handled by lower-paid workers overseas. And some companies see remote human labor as a temporary way to train AI algorithms that will eventually replace them.

For now, people are essential because of the limits of AI and automation. AI has made impressive leaps over the past decade thanks to cheaper and more capable hardware and progress in machine learning, which lets computers learn from examples. But the messy physical world still often confounds algorithms. Robotic arms struggle to pick up and manipulate objects they’ve never seen before, for instance, and self-driving cars cannot make sense of scenes that differ from those in their training data.

Some companies see remote human helpers as a stopgap. This week Ocado, a UK-based retail and technology company, revealed new robots capable of picking items from bins more efficiently. Ocado is one of several companies that use machine learning to train robots to grasp objects from bins. The AI can struggle, because the items are ever changing, making it hard to know how to grab one out of a pile. So human operators help show the robots how to grab when they get stuck. James Matthews, CEO of Ocado’s technology division, said at a Wednesday event that the company has teams of robotic pilots in Mexico and the Philippines who can remotely suggest the best way to grasp new products. The lessons help the algorithms improve, perhaps to the point when humans may not be required.

ArcBest, which uses Phantom Auto’s technology, has developed AI software that lets its forklifts perform some operations on their own, like driving from one point to another. Still, people are needed for more complicated work such as unloading trucks or stacking pallets.


Phantom Auto has sold its remote driving technology to companies working on self-driving cars and delivery robots. These vehicles can handle many situations on the road autonomously but struggle with unusual situations, poor weather conditions, or instrument anomalies, requiring a person to step in and help.

Another company that is hiring remote workers for traditionally physical jobs is Einride, a Swedish maker of self-driving trucks. Currently, Einride’s remote drivers operate only in limited situations, such as moving goods around a large factory in Kentucky that makes General Electric appliances. But CEO Robert Falck says there’s no reason why the drivers couldn’t help drive trucks on real roads. “It's been very obvious that you need to combine remote control with autonomous operation to create a robust and reliable transport setup,” Falck says.

The range of remote work offered on job boards appears to be growing, especially in sectors where AI is being adopted. A number of companies that use drones to inspect or scan work sites or farmland also rely on a combination of autonomy and remote operation.

Travis Deyle, CEO of Cobalt Robotics, which sells robotic security guards to patrol office buildings, says humans are a small but important part of the setup. The robots can navigate around a building autonomously, but humans help when they get stuck or come across something odd. Remote human operators can speak through a microphone if the robot comes across a stranger.

Human teleoperation is “a thinly veiled trade secret” for companies that are working on advanced robotics and automation, says Beane of UCSB. He says that as AI improves, remote operators will no longer be needed for some tasks but says new remote tasks may emerge in other areas.

Beane says one of his research projects shows that the way companies choose to design these roles may make them either dull and simple or interesting and more skilled. He believes this has implications for the future opportunities of those workers. “People have in mind this exploitative picture, but that doesn’t mean it’s necessary,” he says.


Post a Comment