{"id":115044,"date":"2025-03-15T05:45:00","date_gmt":"2025-03-15T09:45:00","guid":{"rendered":"https:\/\/www.freethink.com\/?post_type=ftm_article&p=115044"},"modified":"2025-04-04T13:25:56","modified_gmt":"2025-04-04T17:25:56","slug":"humanoid-1x","status":"publish","type":"ftm_article","link":"https:\/\/www.freethink.com\/artificial-intelligence\/humanoid-1x","title":{"rendered":"Humanoid helpers are now entering our homes"},"content":{"rendered":"\n

It\u2019s 2027. You just introduced yourself to NEO, the humanoid robot who will be living with you for the foreseeable future. The robot is designed to help out with chores and other tasks, and it\u2019s not long before you find yourself grateful not just for its assistance, but also its company.<\/p>\n\n\n\n

Home robots<\/strong><\/h2>\n\n\n\n

Your days of doing chores may be numbered.<\/p>\n\n\n\n

More than 60 years after \u201cThe Jetsons\u201d first introduced the world to the concept of a robot housekeeper, one startup is ready to actually send its humanoid helpers into people\u2019s homes. The robot won\u2019t be wearing Rosie\u2019s French maid uniform, but it could make your life easier\u2014and maybe even improve your mental health.<\/p>\n\n\n\n

Where we\u2019ve been<\/h3>\n\n\n\n

Robots are generally defined as machines that have three things: sensors, programs, and actuators. They can perceive something<\/em> about the world around them, process the information, and then do something physical in response to it.  <\/p>\n\n\n\n

Such systems have been a part of the workforce since the 1960s, when large robotic arms were first integrated into automotive assembly lines. They\u2019ve also been a part of our homes for decades\u2014roombas, washing machines, and even some litter boxes<\/a> meet the \u201crobot\u201d brief.<\/p>\n\n\n\n

Whether they\u2019re in factories or your laundry room, though, these robots all tend to have one thing in common: they\u2019re specialized. They may be able to weld auto parts or clean your clothes, but that\u2019s all they can do. <\/p>\n\n\n\n

A general-purpose robot\u2014one that can perform many<\/em> tasks\u2014is a holy grail of robotics because it would mean you could replace all of these single-purpose bots with just one machine. <\/p>\n\n\n\n

No surprise, then, that so<\/a> many<\/a> groups<\/a> are trying to develop these coveted robots, and because we\u2019ve built our homes, workplaces, and tools around the form and functions of the human body, the majority of developers are doing the same with their robots, giving them human features like legs and arms.<\/p>\n\n\n\n

\"Lineup
Freethink \/ Jacob Hege<\/figcaption><\/div><\/figure>\n\n\n\n

Our bodies are incredible machines, though\u2014your hand alone has 27 degrees of freedom<\/a> and tens of thousands of touch sensors, while your brain is a more efficient processor<\/a> than our best supercomputers\u2014and there\u2019s seemingly no limit to the number of tasks we can perform thanks to this combination of biological hardware and software.<\/p>\n\n\n\n

Not only is replicating even a fraction of our capabilities in a robot a huge engineering challenge, we also need to be able to create these bots at a price that makes economic sense\u2014there\u2019s little incentive to buy or lease one when you could just hire a human or continue to do your chores yourself.<\/p>\n\n\n\n

A number of factors, including advances in artificial intelligence, falling manufacturing costs, and cheaper, better sensors and batteries, now have us at the point where humanoids<\/a> are useful, safe, and cost-effective enough for limited deployment outside of the lab, starting with our workplaces<\/a> and now our homes.<\/p>\n\n\n\n

Where we\u2019re going (maybe)<\/h3>\n\n\n\n

1X Technologies is one of the humanoid developers making that leap into the home. <\/p>\n\n\n\n

CEO Bernt \u00d8yvind B\u00f8rnich founded the company as Halodi Robotics in 2014, and by 2022, its wheeled humanoid robot Eve<\/a> was ready to enter the workforce for logistics and security applications. <\/p>\n\n\n\n

The following year, Halodi changed its name to 1X Technologies and secured another $23.5 million in funding, with OpenAI leading the round. This money would go toward scaling deployment of Eve and developing NEO, a bipedal humanoid designed for home use.<\/p>\n\n\n\n

In August 2024, 1X unveiled<\/a> the first version of that robot, NEO Beta, and in February 2025\u2014a month after announcing that it had raised another $100 million<\/a>\u2014the company announced the next iteration: NEO Gamma<\/a>.<\/p>\n\n\n\n

\"Humanoid
1X Technologies<\/figcaption>
NEO Gamma\n<\/div><\/div><\/figure>\n\n\n\n

In the six months between the debuts of Beta and Gamma, 1X upgraded several facets of NEO, making the robot quieter, more reliable, and generally safer and more pleasant to be around. 1X is now looking for people who are interested in sharing their homes with the humanoid.<\/p>\n\n\n\n

These early adopters should get some use out of the robot, according to B\u00f8rnich, but the main goal of the deployment is to collect the kind of training data needed to take humanoid development to the next level.  <\/p>\n\n\n\n

\u201cThere’s very little secret sauce in the AI space,\u201d he tells Freethink. \u201cA lot of people think that it’s about some very novel breakthroughs in how you train these algorithms, but it’s not. It’s all about the data, and if your data is better than everyone else’s, your model will be better.\u201d <\/p>\n\n\n\n

ChatGPT is a prime example. Because OpenAI had access to a massive dataset of written content for training\u2014the internet\u2014it was able to develop an AI model with an unprecedented ability to interpret and generate text.<\/p>\n\n\n\n

When it comes to training an AI to use a robot body to help us out around the house, though, there is no such massive dataset of examples. Developers can show a model relevant YouTube videos and have it train in computer simulations, but that isn\u2019t enough to teach it how to go into any home and do everything we want it to be able to do.<\/p>\n\n\n\n

“Intelligence comes from diversity of thought. If you only see one thing every day, you’re not going to get very smart.”<\/p>Bernt \u00d8yvind B\u00f8rnich<\/cite><\/blockquote><\/figure>\n\n\n\n

To get to that point, we are going to need a lot more training data, and according to B\u00f8rnich, the best way to get it is to send NEO Gamma into a variety of homes and have it perform useful tasks, sometimes with a human teleoperating the bot and sometimes with the AI controlling it. <\/p>\n\n\n\n

\u201cIntelligence comes from diversity of thought,\u201d says B\u00f8rnich. \u201cIf you only see one thing every day, you’re not going to get very smart. You need to see a lot of different scenarios.\u201d<\/p>\n\n\n\n

\u201cI like to say if you want to be good at something as simple as picking up a coffee cup, you want to see all the coffee cups in the world in as [many] different settings among people as possible,\u201d he continues. <\/p>\n\n\n\n

1X kicked off this home deployment phase in September 2024, sending a NEO Gamma to live with filmmaker Jason Carman for two days. During its visit, the robot\u2014operated by a human remotely\u2014made Carman a cup of coffee, told a dad joke, and even helped with the cooking.<\/p>\n\n\n\n

\u201cHe picked up the egg and handed it to me, and it felt very monumental, I guess,\u201d Carman says<\/a> in a video documenting the experience. \u201cIt was a very strange feeling because, obviously, we’re filming this for an episode, there’s people with cameras and some of the engineers from 1X were standing by, but I don’t know. It felt like a sci-fi moment becoming reality.\u201d<\/p>\n\n\n\n

\n

“It’s absolutely unforgivable to have any breach of privacy in these kinds of systems.”<\/p>\nBernt \u00d8yvind B\u00f8rnich<\/cite><\/blockquote>\n\n\n\n

As 1X deploys more and more NEOs, the amount of training data the robots collect will grow until it dwarfs the amount of data coming from the internet, according to B\u00f8rnich: \u201cThe data pyramid will flip where almost all data comes from robots and the internet is a small part of it.\u201d<\/p>\n\n\n\n

Because many people would, understandably, be wary of having a robot walking around their home, recording everything it sees and hears, 1X has made protecting customers\u2019 privacy a top priority in NEO\u2019s development.<\/p>\n\n\n\n

\u201cHow do you ensure that this is Big Sister and not Big Brother?\u201d says B\u00f8rnich. \u201cI love that analogy because we need your data. If we’re going to get an abundance of labor, there is just one way: We need your data.\u201d<\/p>\n\n\n\n

\u201cBut we need to do this in a manner where we respect that and respect your privacy to the extreme,\u201d he continues. \u201cIt’s absolutely unforgivable to have any breach of privacy in these kinds of systems.\u201d<\/p>\n\n\n\n

\"A
1X Technologies<\/figcaption>
A NEO Gamma helping with breakfast\n<\/div><\/div><\/figure>\n\n\n\n

To that end, 1X has designed NEO Gamma so that no one at the company, no matter how high their access level, will be able to see the data it collects in customers\u2019 homes\u2014the model<\/em> can see it and train on it, but that\u2019s it. <\/p>\n\n\n\n

It has also implemented a time delay between when data is recorded and when it is uploaded. This will give customers time to delete recordings if the robot ever sees something they aren\u2019t comfortable having included in its training set.<\/p>\n\n\n\n

If someone at 1X is interested in analyzing data collected by a NEO\u2014to figure out why the robot did a certain thing, for example\u2014they will send a request to the customer via their phone app. That person will then have an opportunity to see a preview of the data before deciding whether or not to let 1X see it.<\/p>\n\n\n\n

\u201cYou need to give permission, and if you give permission explicitly, the one specific person in question at our company who asked for it can see the data, but no one else,\u201d says B\u00f8rnich.<\/p>\n\n\n\n

\n