If you could have a robot at home, what would it do?
In the case of the 1985 sitcom Small Wonder, it would be a member of the family named Vicki. But pop culture has more frequently envisioned robots as highly competent helpers, such as Rosey the housekeeper in the Jetsons and the super-savvy AI assistant Jarvis in the Iron Man and Avengers movies.
Science fiction writer Joanna Maciejewska captured that sensibility in a pithy post on X (formerly Twitter) earlier this year: “I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes.”
She hit a nerve. Her post has racked up 3 million views and 102,000 likes since it was posted on March 29.
It’s a lifestyle The Jetsons envisioned way back in 1962 — but, 62 years later, we’re still not there yet.
Why?
For starters, despite decades of research into AI and robotics, it remains a formidable technical challenge to integrate the technology into our lives logically and affordably. There are philosophical and ethical considerations as well. As one response to Maciejewska pointed out, it’s a complicated topic:
“Who decides what we love to do and what needs to be automated?” the respondent wrote. “I work with a ton of accountants concerned with AI taking their job but [who] love the idea of AI helping them write articles.”
We’re all thinking and talking about this a lot right now because of the advent of generative AI, which, among other things, has shown a flair for writing, if not so much with math (or taxes). ChatGPT, Google Gemini, Microsoft Copilot, Meta AI, Adobe Firefly and many other AI chatbot systems are remarkable for their ability to synthesize and process language and images in a very humanlike way.
But chatbots are not robots.
Maciejewska herself clarified that she’s not looking for an actual laundry robot, but hoping AI will take on tasks she hates, like doing taxes. (Another X user noted that AI has “no capacity for understanding or judgment and therefore cannot be trusted with actual tasks like taxes.”) She didn’t respond to a request for comment for this article.
And yet the dream of laundry robots persists. So it’s important to understand both the potential and the limitations and how technology developers are investing their time and energy in this work in progress.
AI versus robotics
There’s an important distinction between AI and robotics. (Just ask folks on Reddit.)
AI is the branch of computer science focused on simulated human intelligence in machines. You can think of it as the brains behind the operation. It’s the software, essentially.
Robotics, on the other hand, refers to machines: physical, mechanical things that can do jobs like pouring a glass of wine, putting dirty dishes in a dishwasher, placing a flower in a vase or making an omelet. Or, in factories rather than homes, assembling cars and shuttling products in a warehouse.
Watch this: Nvidia’s Project GR00T vs. Tesla Optimus: Competing Robot Strategies
This is harder to pull off because it requires hardware — some kind of body with arms and hands that can interact with the world around it and manipulate objects of different sizes and textures. It’s lagging far behind the explosive growth in generative AI because, well, it’s really hard to build a robot that can do all the physical things the human body can do and that can understand the surrounding environment.
Even when we see a robot that can perform one particular task really well — like brew beer, make ice cream or cook ramen — that’s it. And while singular use cases make for flashy examples at trade shows, they’re often a harder sell to consumers like you and me — pricey novelties, if they ever actually make it to market.
Laundry (and dishes)
Let’s consider the laundry example for a moment.
Laundry requires physical manipulation of objects in a complicated environment: your laundry room. There’s the washer and the dryer and potentially the ironing board as well. Shirts, pants, blouses, bras, socks, towels, sheets and more with different sizes, shapes, textures and cleaning requirements.
It’s a challenge that researchers and startups have put their minds and their dollars toward, without much to show for it.
Back in 2010, for instance, robotics company Willow Garage hoped that by giving away $4 million worth of robots to 11 researchers, it could advance the state of general-purpose robots. (The headline on CNET’s story 14 years ago? “Getting robots to do the laundry and the dishes.”) While one graduate student made progress on folding towels, the startup shut down in 2014 after reportedly spinning off most of its innovations into other private ventures.
Here’s what Foldimate could do, back in 2019.
A few years later, a startup called Foldimate debuted at CES with a $980 laundry-folding robot, but you still had to feed each article individually into the machine, and it couldn’t handle items like sheets, towels or baby clothes. It ceased to exist circa 2020.
Sometimes it’s not even a robotics issue. Amazon and Whirlpool tried to automate the process of ordering detergent with a smart washing machine they showcased at CES 2016. By linking your Amazon account to the Whirlpool app, you could give the machine permission to reorder your favorite detergent when it was running low.
Whirlpool and Amazon didn’t respond to requests for comment about the fate of that particular washing machine, but it’s safe to say the concept didn’t catch on.
Building and training a household robot
Meanwhile, the work goes on, trying to crack the nut that is household robots — or maybe better said, robots that can do chores of any kind. And that requires not just a physical robotics component but training to simulate human intelligence.
Researchers at Stanford have developed a robot called Mobile Aloha, which can autonomously put away a cooking pot, push in chairs, sauté shrimp, clean up a wine spill and give high-fives. You can see it — complete with googly eyes — prepare a three-course meal here. A release from the university in April said Mobile Aloha has also shown promise in additional household tasks like vacuuming, doing laundry and watering plants. But it’s a long way from general availability, and in its current form, well, let’s just say it’s not exactly sleek.
The Mobile Aloha team is training the robot through a process known as imitation learning. Here, a human stands behind the robot and uses a teleoperation interface to show the robot how to complete a task with its own arms. Chelsea Finn, assistant professor of computer science and electrical engineering at Stanford and adviser to the Mobile Aloha team, in a demonstration video described this process as “kind of like a puppeteering setup” with the robot as the puppet.
It takes about 50 tries per task to yield enough data for the robot, Tony Zhao, a graduate student in computer science at Stanford and co-lead of the Mobile Aloha team, said in the video. The ultimate goal is to “appeal to what people think a future home robot should look like,” he added.
To date, most robots have been in controlled environments like factories or warehouses where they can be programmed to perform the same motion repeatedly. To a degree, that will include the use of AI techniques such as machine learning.
“For robots to be successful out in the real world, they need to be able to perceive their surroundings and react to their surroundings,” Finn said. “We’re interested in seeing whether we can leverage machine learning to allow robots to be more intelligent and actually push them out into the real world.”
And while the robot has shown promise in autonomously performing a range of household tasks, it’s a $32,000 prototype at this stage.
The likes of Google, Amazon, Apple and Tesla have their own projects and prototypes in the works, designed to tackle, in one way or another, a variety of tasks from cleaning to monitoring homes to performing unsafe, repetitive or boring tasks.
In a 2022 demonstration of the rather terrifyingly humanoid Tesla Bot, CEO Elon Musk said he hoped it would go on sale by 2027 for $20,000, but the bots will first be put to work in Tesla’s factories, where potential jobs could include carrying parts to other robots on the manufacturing line. Personal butler? Not anytime soon.
The first prototype of Optimus, the code name for the Tesla Bot, waves an arm at the Tesla AI Day 2022 event.
‘A tool in our hands’
There’s still a long way to go before we combine both AI and robotics in a machine that can perform multiple tasks for us in the unpredictable environments that are our homes. And whatever it’s that Mobile Aloha, Tesla Bot and omelet flippers can do, they can’t write poetry or help you finish a cover letter or generate images of teddy bears and kittens at the beach.
Yet here we are, with generative AI tools spreading like wildfire as the companies behind them push them out into the marketplace. We haven’t had tools like these before, with so many potential uses that we’re just starting to explore — including in creative fields that feel especially human — and with such a low barrier to entry in terms of cost.
“If I’m developing AI for creative tasks, it’s going to be a far more lucrative investment with potential applications in advertising, entertainment, design and other super high-value industries,” said Jason Alan Snyder, global chief technology officer at advertising agency Momentum Worldwide.
Meanwhile, laundry and kitchen automation are well-established and stable, given decades of refinements already applied to them. Washing machines have been around since the 1850s and dishwashers since the 1950s.
Occasionally, we do see a breakthrough advance in household technology. In 2002, iRobot released its first robot vacuum, and it has since sold more than 40 million devices. Robot lawn mowers are just now starting to catch on.
And so perhaps we just need to be a little more patient to allow the field of home robots to evolve.
Meanwhile, maybe the underlying, worrisome issue isn’t so much about the specific tasks being done as it is about our role in the world.
Karin Kimbrough, chief economist at LinkedIn, said Maciejewska’s X post really speaks to a broad anxiety over whether AI is going to make us less creative — and perhaps destroy the creative process altogether.
“I would say no, not necessarily,” she said, pointing to an earlier fear about calculators taking away our math skills. “It just became a tool to do some math faster.”
Generative AI can help us create content, but we don’t have to use it, Kimbrough noted, adding that there are many ways people can express their creativity. And even though these tools can help us do some work faster, they still require instruction and oversight, like fact-checking and editing.
“It is a tool in our hands,” Kimbrough said.
