A robot hand wiping some egg off its fingers with a towel is the most bizarrely human thing you'll see a robot do all week, and it's just the beginning
A glove, some cameras, and a very special robot hand
- Genesis AI robot hands handle frontier tasks
- They can make eggs and solve a Rubik's Cube
- The training is done, in part, via low-cost, wearable gloves
If there's a single hurdle preventing humanoid robots from entering our homes and workplaces, it might be "frontier tasks". These are complex, multi-step tasks like operating as a lab assistant, solving a Rubik's Cube, or making a delicious smoothie or omelet. They're the kind of things most humans can do without thinking, but for robots and AI, it's darn near impossible to match your average lab tech or short-order cook at these skills.
Today, though, I watched a robot casually wipe some egg yolk off its fingers as it prepped some lightly scrambled eggs. It was such a normal thing to do in the course of successfully cooking a meal that, for a second, I forgot I was watching some disembodied robot hands accomplish the task.

Those same highly dexterous robot hands, all built and programmed by Genesis AI, also solved a Rubik's Cube, did more than passable lab assistant work, and even made a delicious-looking purple smoothie. The hands, part of a three-part system for AI development, training, and deployment, look set to change how we think about robots at home and in the workplace.
Article continues below"I don't know, if you know this," Vivian Sun, Genesis AI's Vice President of Commercial and Strategy, told me, "80% of human labor are doing it with their hands." This, it turns out, presents a huge problem with our long-dreamed-of and fast-approaching robotics future. The kind of work we do with our hands takes human-level (which is considerable) patience, skill, and dexterity.
Making and training a believable robot hand
Solving that problem and delivering a solution to its Fortune 500 customers who have a multitude of use cases for such robots became a priority for Genesis AI. The solution was not just a robot or even an algorithm. It's an end-to-end solution that starts with data collection and a set of proprietary gloves.
Trainers wear the thin, wireless gloves, which are covered in sensors and even include a camera, while they perform these frontier tasks, like making an egg, hundreds of times. "[It's] the most natural way to interact with the physical world," explained Sun.
Those same trainers are also wearing a head-mounted camera to watch how, for instance, a human breaks an egg.
Sign up for breaking news, reviews, opinion, top tech deals, and more.
That training data is then combined with "internet-level data", basically thousands of online videos showing people breaking eggs and making scrambled eggs.
It's all valuable data fed into Genesis's new Gene foundation model version 26.5, and that is then used to train and drive Genesis AI's proprietary robot hands.
Look and work like the real thing
It's unlikely Genesis AI would have the same kind of success if they were to feed this intricate data into a pair of claws.
"We have the most human-like robotic…hand. These hands are made proprietary to Genesis AI...[they] look exactly like a human hand in terms of the functionality, in terms of the proportions, in terms of the size and shapes," said Sun.
They close a critical domain gap not just by looking like human hands but, as I noticed in the videos, moving just like them, too. In the egg video, under a pair of latex gloves, you could mistake them for human palms and 10 digits, except for the robot trunk they're attached to.
The end-to-end process allows for a virtually direct transfer of human skill to robot embodiment. It is, Geneis AI believes, a scalable solution. Sun told me that the training glove set is far more affordable and sensible than the current robotic training method: teleoperation.
Teleoperation "is quite bulky. You have to stage the setup, hire the people... and most of the time you can’t even recreate that scenario. For example, if you're trying to learn jet engine repair, how would you build a jet engine within your facility? That’s just impossible," said Sun.
She couldn't quote a price for the gloves or robot hands, but told me the gloves will be "50 times less expensive than the next thing within the industry," and added that it's 100 times cheaper than teleoperation.
What's next
Long-term, Genesis AI envisions mass-producing the glove and sending it to businesses and homes, where wearers can help train up Genesis AI robots.
Of course, that's just a pair of highly articulate hands right now, but Sun told me they are building a complete robot.
Would they call it "Genesis Robot"? Sun wouldn't say, but laughed, "We have a name that we will tell you very soon. We went through multiple workshops coming out of the name. So yes, there is a name."
For now, though, we have this small collection of frontier task videos that highlight what may be the apex of robot dexterity. I asked Sun what she thought the first time she viewed them.
"I was personally just stunned....seeing it in reality, performing this type of long-horizon, very complicated tasks, it was just like almost like a moment of reality check: Here is the world we're living in."
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course, you can also follow TechRadar on YouTube and TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

A 38-year industry veteran and award-winning journalist, Lance has covered technology since PCs were the size of suitcases and “on line” meant “waiting.” He’s a former Lifewire Editor-in-Chief, Mashable Editor-in-Chief, and, before that, Editor in Chief of PCMag.com and Senior Vice President of Content for Ziff Davis, Inc. He also wrote a popular, weekly tech column for Medium called The Upgrade.
Lance Ulanoff makes frequent appearances on national, international, and local news programs including Live with Kelly and Mark, the Today Show, Good Morning America, CNBC, CNN, and the BBC.
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.