A Robot in Your Kitchen?

Neo is a humanoid butler that works in Redwood City, California. Clad in a beige bodysuit, the robot greets you at the door with a handshake.

“I have a firm grip,” it says, its glassy black eyes—camera lenses—staring out from behind a reflective faceplate.

When the home’s owner, engineer Bernt Børnich, asks for a bottle of water, Neo walks into the kitchen and gets one from the fridge. Later Neo will do the laundry.

Artificial intelligence is already driving cars, writing essays, and even generating computer code. Now machines built to look like humans and powered by A.I. are poised to move into our homes to help with the chores.

Neo is a robot butler that works in Redwood City, California. The humanoid wears a beige bodysuit and greets you at the door with a handshake.

“I have a firm grip,” it says, its glassy black eyes—camera lenses—looking out from behind a reflective faceplate.

When the home’s owner, engineer Bernt Børnich, asks for a bottle of water, Neo walks into the kitchen and gets one from the fridge. Later Neo will do the laundry.

Artificial intelligence is already doing human tasks. It is driving cars, writing essays, and even generating computer code. Now machines built to look like humans and powered by A.I. are almost ready to move into our homes to help with the chores.

But not quite yet. Neo is operated by a technician in Børnich’s basement as part of an effort to teach it how to navigate the real world.

Børnich’s start-up, 1X, is among the dozens of companies, including the electric carmaker Tesla, that want to sell humanoids and get them into homes and businesses. Børnich says these robots—which mimic the ways people walk, bend, twist, reach, and grip—will one day do much of the physical work now handled by people, such as emptying dishwashers, sorting packages in warehouses, or using tools.

But not quite yet. Neo is operated by a technician in Børnich’s basement. The technician is working to teach it how to navigate the real world.

Børnich’s start-up, 1X, is among the dozens of companies, including the electric carmaker Tesla, that want to sell humanoids. They hope to get them into homes and businesses. These robots mimic the ways people walk, bend, twist, reach, and grip. Børnich says they will one day do much of the physical work now handled by people, such as emptying dishwashers, sorting packages in warehouses, or using tools.

The push toward humanoid labor is being fueled by advances in robotic hardware and A.I. that allow robots to rapidly learn new skills.

“There are a lot of people like me,” Børnich says. “They’ve dreamed of having something like this in their home since they were a kid.”

But it could be years before these machines are fully up to the task. Most of them have to be remotely guided. When Neo speaks, it’s the technician’s voice you hear. Neo can walk through the dining room and kitchen on its own, but the technician has to guide its hands via a virtual reality headset and two wireless joysticks. 1X hopes these robots will soon operate in homes autonomously.

Børnich and his business partner, robotics engineer Eric Jang, first started training Neo to walk in a digital environment. By simulating the physics of the real world in a video-game-like setting, they created a digital version of their robot that could stand, balance, and eventually take steps. Then they transferred everything the virtual robot had learned to a physical humanoid.

They’re making progress. Neo can step around objects that move into its path and, if bumped, stay on its feet.

“All of this is learned behavior,” Jang says. “If we put it into any environment, it should know how to do this.”

The push toward humanoid labor is being fueled by advances in robotic hardware.  Developments in A.I. now allow robots to rapidly learn new skills.

“There are a lot of people like me,” Børnich says. “They’ve dreamed of having something like this in their home since they were a kid.”

But it could be years before these machines are fully ready. Most of them have to be remotely guided by a person. It’s the technician’s voice you hear when Neo speaks. Neo can walk through the dining room and kitchen on its own. But the technician still has to guide its hands via a virtual reality headset and two wireless joysticks. 1X hopes these robots will soon operate in homes independently.

Børnich and his business partner, robotics engineer Eric Jang, first started training Neo to walk in a digital environment. They simulated the physics of the real world in a video-game-like setting. Then they created a digital version of their robot that could stand, balance, and eventually take steps. Once the virtual robot could move, they transferred everything the virtual robot had learned to a physical humanoid.

They’re making progress. Neo can step around objects that move into its path. It can stay on its feet if bumped.

“All of this is learned behavior,” Jang says. “If we put it into any environment, it should know how to do this.”

But the physics of loading a dishwasher or folding laundry are exceedingly complex, and 1X can’t teach these tasks in the virtual world. It has to gather data in real environments.

At Børnich’s home, Neo struggles with the refrigerator door when the Wi-Fi flickers. Even when controlled by a technician, Neo might drop a cup or have a hard time tossing an empty bottle into a garbage can under a sink.

Yet just as ChatGPT can learn to write term papers by analyzing text culled from the internet, a robot can learn to clean windows by pinpointing patterns in hours of digital video. As they guide Neo through household chores, Børnich and his team gather this data, using cameras and sensors installed on the robot itself.

By selling humanoids into homes, 1X hopes it can gather enormous amounts of data. That is, if it can find buyers: Building Neo costs about as much as manufacturing a small car—tens of thousands of dollars. Buyers might be skeptical. Though many Americans have grown used to asking chatbots for directions or restaurant recommendations, would they feel comfortable sharing their home with one? And would the androids live up to the science fiction hype?

“There are many videos out there that give a false impression of these robots,” says Ken Goldberg, a robotics professor at the University of California, Berkeley. “Though they look like humans, they aren’t always behaving like humans.”

But the physics of loading a dishwasher or folding laundry are extremely complex. These tasks can’t be taught in the virtual world. It has to be learned by gathering data in real environments.

At Børnich’s home, Neo struggles with the refrigerator door when the Wi-Fi flickers. Even when controlled by a technician, Neo might drop a cup. It might have a hard time tossing an empty bottle into a garbage can under a sink.

A robot can learn to clean windows by pinpointing patterns in hours of digital video. As they guide Neo through household chores, Børnich and his team use cameras and sensors installed on the robot to gather data.

If 1X can find buyers for home humanoids, they hope to gather enormous amounts of data.  But building Neo costs tens of thousands of dollars, about the cost of manufacturing a small car. Buyers might be skeptical. Americans have grown used to asking chatbots for directions or restaurant recommendations. But would they feel comfortable sharing their home with one? And would the androids live up to the science fiction hype?

“There are many videos out there that give a false impression of these robots,” says Ken Goldberg, a robotics professor at the University of California, Berkeley. “Though they look like humans, they aren’t always behaving like humans.”

But with rapid advancements in A.I., robots like Neo could become viable sooner rather than later. That would continue a decades-long trend of machines taking over repetitive, hands-on jobs, like those in factories and on assembly lines. As for other kinds of vocations requiring creativity and decision-making skills, such as those of doctors and social workers, robots likely won’t replace humans, many experts say—at least not anytime soon.

They might, however, help ease growing shortages in some employment sectors, including, for example, workers who handle both housecleaning and care of elders and children. Organizations that represent people in those professions welcome new technologies—provided they can perform well alongside human workers and not supplant them entirely.

“These tools could make some of the more strenuous, taxing, and dangerous work easier, and allow workers to focus on things that only human workers can offer,” says Ai-jen Poo, president of the National Domestic Workers Alliance.

Whether people will let these machines into their homes, even if robots develop the necessary skills, is an open question. In the meantime, Børnich is pressing forward with Neo, one dirty dinner plate at a time. 

But with rapid advancements in A.I., robots like Neo could become possible sooner than later. There has been a decades-long trend of machines taking over repetitive, hands-on jobs, like those in factories and on assembly lines. As for other kinds of jobs requiring creativity and decision-making skills, such as those of doctors and social workers, many experts say robots likely won’t replace humans anytime soon.

They might, however, help ease growing shortages in some employment sectors, such as workers who handle both housecleaning and care of elders and children. Organizations that represent people in those professions welcome new technologies as long as they can perform well alongside human workers and not replace them entirely.

“These tools could make some of the more strenuous, taxing, and dangerous work easier, and allow workers to focus on things that only human workers can offer,” says Ai-jen Poo, president of the National Domestic Workers Alliance.

It is still an open question if people will let these machines into their homes, even if robots develop the necessary skills. In the meantime, Børnich is pressing forward with Neo, one dirty dinner plate at a time.

Cade Metz is a technology reporter for The New York Times.

Cade Metz is a technology reporter for The New York Times.