Tech
This Robot Only Needs a Single AI Model to Master Humanlike Movements

While there is a lot of work to do, Tedrake says all of the evidence so far suggests that the approaches used to LLMs also work for robots. “I think it’s changing everything,” he says.
Gauging progress in robotics has become more challenging of late, of course, with videoclips showing commercial humanoids performing complex chores, like loading refrigerators or taking out the trash with seeming ease. YouTube clips can be deceptive, though, and humanoid robots tend to be either teleoperated, carefully programmed in advance, or trained to do a single task in very controlled conditions.
The new Atlas work is a big sign that robots are starting to experience the kind of equivalent advances in robotics that eventually led to the general language models that gave us ChatGPT in the field of generative AI. Eventually, such progress could give us robots that are able to operate in a wide range of messy environments with ease and are able to rapidly learn new skills—from welding pipes to making espressos—without extensive retraining.
“It’s definitely a step forward,” says Ken Goldberg, a roboticist at UC Berkeley who receives some funding from TRI but was not involved with the Atlas work. “The coordination of legs and arms is a big deal.”
Goldberg says, however, that the idea of emergent robot behavior should be treated carefully. Just as the surprising abilities of large language models can sometimes be traced to examples included in their training data, he says that robots may demonstrate skills that seem more novel than they really are. He adds that it is helpful to know details about how often a robot succeeds and in what ways it fails during experiments. TRI has previously been transparent with the work it’s done on LBMs and may well release more data on the new model.
Whether simple scaling up the data used to train robot models will unlock ever-more emergent behavior remains an open question. At a debate held in May at the International Conference on Robotics and Automation in Atlanta, Goldberg and others cautioned that engineering methods will also play an important role going forward.
Tedrake, for one, is convinced that robotics is nearing an inflection point—one that will enable more real-world use of humanoids and other robots. “I think we need to put these robots out of the world and start doing real work,” he says.
What do you think of Atlas’ new skills? And do you think that we are headed for a ChatGPT-style breakthrough in robotics? Let me know your thoughts on ailab@wired.com.
This is an edition of Will Knight’s AI Lab newsletter. Read previous newsletters here.
Tech
Europe’s fastest supercomputer to boost AI drive

Europe’s fastest supercomputer Jupiter is set to be inaugurated Friday in Germany with its operators hoping it can help the continent in everything from climate research to catching up in the artificial intelligence race.
Here is all you need to know about the system, which boasts the power of around one million smartphones.
What is the Jupiter supercomputer?
Based at Juelich Supercomputing Center in western Germany, it is Europe’s first “exascale” supercomputer—meaning it will be able to perform at least one quintillion (or one billion billion) calculations per second.
The United States already has three such computers, all operated by the Department of Energy.
Jupiter is housed in a center covering some 3,600 meters (38,000 square feet)—about half the size of a football pitch—containing racks of processors, and packed with about 24,000 Nvidia chips, which are favored by the AI industry.
Half the 500 million euros ($580 million) to develop and run the system over the next few years comes from the European Union and the rest from Germany.
Its vast computing power can be accessed by researchers across numerous fields as well as companies for purposes such as training AI models.
“Jupiter is a leap forward in the performance of computing in Europe,” Thomas Lippert, head of the Juelich center, told AFP, adding that it was 20 times more powerful than any other computer in Germany.
How can it help Europe in the AI race?
Lippert said Jupiter is the first supercomputer that could be considered internationally competitive for training AI models in Europe, which has lagged behind the US and China in the sector.
According to a Stanford University report released earlier this year, US-based institutions produced 40 “notable” AI models—meaning those regarded as particularly influential—in 2024, compared to 15 for China and just three for Europe.
“It is the biggest artificial intelligence machine in Europe,” Emmanuel Le Roux, head of advanced computing at Eviden, a subsidiary of French tech giant Atos, told AFP.
A consortium consisting of Eviden and German group ParTec built Jupiter.
Jose Maria Cela, senior researcher at the Barcelona Supercomputing Center, said the new system was “very significant” for efforts to train AI models in Europe.
“The larger the computer, the better the model that you develop with artificial intelligence,” he told AFP.
Large language models (LLMs) are trained on vast amounts of text and used in generative AI chatbots such as OpenAI’s ChatGPT and Google’s Gemini.
Nevertheless with Jupiter packed full of Nvidia chips, it is still heavily reliant on US tech.
The dominance of the US tech sector has become a source of growing concern as US-Europe relations have soured.
What else can the computer be used for?
Jupiter has a wide range of other potential uses beyond training AI models.
Researchers want to use it to create more detailed, long-term climate forecasts that they hope can more accurately predict the likelihood of extreme weather events such as heat waves.
Le Roux said that current models can simulate climate change over the next decade.
“With Jupiter, scientists believe they will be able to forecast up to at least 30 years, and in some models, perhaps even up to 100 years,” he added.
Others hope to simulate processes in the brain more realistically, research that could be useful in areas such as developing drugs to combat diseases like Alzheimer’s.
It can also be used for research related to the energy transition, for instance by simulating air flows around wind turbines to optimize their design.
Does Jupiter consume a lot of energy?
Yes, Jupiter will require on average around 11 megawatts of power, according to estimates—equivalent to the energy used to power thousands of homes or a small industrial plant.
But its operators insist that Jupiter is the most energy-efficient among the fastest computer systems in the world.
It uses the latest, most energy-efficient hardware, has water-cooling systems and the waste heat that it generates will be used to heat nearby buildings, according to the Juelich center.
© 2025 AFP
Citation:
Europe’s fastest supercomputer to boost AI drive (2025, September 5)
retrieved 5 September 2025
from https://techxplore.com/news/2025-09-europe-fastest-supercomputer-boost-ai.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
Tech
Lenovo’s New Laptop Concept Can Swivel the Screen From Landscape to Portrait

Lenovo isn’t shy about trying new things. Last year, the PC maker teased a concept laptop with a transparent screen. Earlier this year, the ThinkBook Flip concept employed a flexible OLED display that folded over the top of the laptop lid, ready to flip up whenever you needed the extra screen space. At CES 2025, we saw a ThinkBook with a rollable OLED screen that expanded upward automatically at the touch of a button—this one is a real product you can actually buy.
Get ready for another whacky concept. At IFA 2025, the tech exhibition in Berlin, Lenovo unveiled its latest idea: the Lenovo ThinkBook VertiFlex. This is a laptop with a screen that can manually swivel from a standard horizontal orientation to vertical.
Portrait Mode
By default, the ThinkBook VertiFlex Concept looks like a normal 14-inch laptop. Look closely at the screen’s edge, however, and you’ll see a second layer jutting out; that’s the actual screen. Grab the right corner edge of the screen and push it upward, and the display will smoothly swivel up into a vertical orientation.
The back panel the screen is mounted on has a felt backing to keep everything smooth and scratch-free, and you can even prop a phone up here in this orientation. There’s a mechanism inside that manages the motion and keeps it operating smoothly. Despite this, the PC is still fairly slim at 17.9 mm, and it weighs roughly 3 pounds. (The 14-inch MacBook Pro is around 15 mm thick and weighs 3.4 pounds.)
I use a dual-screen setup with one vertical monitor next to my main ultrawide monitor at home. Having a vertical screen is a game-changer, as it’s perfect for applications that utilize more vertical space. Email is a great example, so are apps like Slack, anything to do with PDFs, and even most word processing software. But I’ve yet to change my screen orientation in the middle of a workflow.
Tech
Get Our Favorite Smart Lock for Just $164 Right Now

Is your current smart lock frustrating you endlessly, like mine is? The Yale Approach Smart Lock (8/10, WIRED Review) is currently marked down to just $164 on Amazon, a healthy 32% discount on our editors’ top pick for smart locks. This sale comes at a perfect time, because I was just complaining about the fingerprint reader on mine no longer working.
The Yale Approach uses part of your existing deadbolt, which is great news for renters who don’t want to make major changes. You’ll also get to use your existing keys to unlock the deadbolt, which can save you a trip to the locksmith. There’s also a wi-fi bridge that needs a nearby plug to provide other services, but that’s not uncommon for smart locks. Our reviewer, Nena Farrell, even said it “works perfectly,” which is great news, because I have to unplug mine and plug it back in at least once a week.
Approach isn’t just a name, as this smart deadbolt’s standout feature is auto-unlock. By setting up your location in the Yale Access App, you can set the bolt to unlock as your get close to home, which our reviewer said “worked smoothly”, as long as she got far enough away from home for it to recognize her return. There’s an auto-lock, too, using timers from 10 seconds to 30 minutes.
This version of the Yale Approach includes the touchscreen keypad, which needs its own flat space to either stick or screw to. In exchange, it lets you set codes for yourself or friends, with options for time and access limits if you need to manage entry to your home more carefully. It also gives you an easy button to press to lock the deadbolt as you leave the house, and a biometric fingerprint scanner.
No matter what smart lock you buy, there’s going to be a little bit of hassle, that just comes with the territory, unfortunately. The Yale smooths out a lot of the worst parts by adapting to your existing hardware, and mostly stays out of the way afterwards. The auto-unlock feature isn’t totally unique to the Approach, but it is currently our favorite implementation. The price is normally a bit on the high side, so the discount here makes this a very appealing pickup for anyone ready to relegate their old front door lock to the garage door, like I’m about to.
-
Sports1 week ago
New Zealand rugby player Shane Christie, who suffered multiple concussions, dies aged 39 – SUCH TV
-
Tech1 week ago
Top CDC Officials Resign After Director Is Pushed Out
-
Fashion1 week ago
Portugal Jewels Chiado boutique nominated for two global design awards
-
Sports1 week ago
New-look Pac-12 extends CW deal through 2031
-
Fashion1 week ago
ICE cotton futures fall for 2nd consecutive day on strong crop outlook
-
Fashion1 week ago
Israel’s Delta Galil posts $470 mn Q2 sales, updates 2025 guidance
-
Sports1 week ago
Dolphins GM Chris Grier says fans threatened his family in string of vile emails after team’s lackluster year
-
Entertainment1 week ago
YouTube TV viewers could lose access to Fox channels over contract dispute