If someone can solve a Rubik’s Cube, you might safely assume they are both nimble-fingered and good at puzzles. That may not be true for a cube-conquering robot.
OpenAI, a research company in San Francisco whose founders include Elon Musk and Sam Altman, made a splash Tuesday by revealing a robotic system that learned to solve a Rubik’s cube using its humanoid hand.
In a press release, OpenAI claimed that its robot, called Dactyl, is “close to human-level dexterity.” And videos of the machine effortlessly turning and spinning the cube certainly seem to suggest as much. The clips were heralded by some on social media as evidence that a revolution in robot manipulation has at long last arrived.
In fact, it may be some time before robots are capable of the kind of manipulation that we humans take for granted.
There are serious caveats with the Dactyl demo. For one thing, the robot dropped the cube eight out of 10 times in testing—hardly evidence of superhuman, or even human, deftness. For another, it required the equivalent of 10,000 years of simulated training to learn to manipulate the cube.
“I wouldn’t say it’s total hype—it’s not,” says Ken Goldberg, a roboticist at UC Berkeley who also uses reinforcement learning, a technique in which artificial intelligence programs “learn” from repeated experimentation. “But people are going to look at that video and think, ‘My God, next it’s going to be shuffling cards and other things,’ which it isn’t.”
Showy demos are now a standard part of the AI business. Companies and universities know that putting on an impressive demo—one that captures the public’s imagination—can produce more headlines than just an academic paper and a press release. This is especially important for companies competing fiercely for research talent, customers, and funding.
Others are more critical of the demo and the hoopla around it. “Do you know any 6-year-old that drops a Rubik’s cube 80 percent of the time?” says Gary Marcus, a cognitive scientist who is critical of AI hype. “You would take them to a neurologist.”
More important, Dactyl’s dexterity is highly specific and constrained. It can adapt to small disturbances (cutely demonstrated in the video by nudging the robot hand with a toy giraffe). But without extensive additional training, the system can’t pick up a cube from a table, manipulate it with a different grip, or grasp and handle another object.
“From the robotics perspective, it’s extraordinary that they were able to get it to work,” says Leslie Pack Kaelbling, a professor at MIT who has previously worked on reinforcement learning. But Kaelbling cautions that the approach likely won’t create general-purpose robots, because it requires so much training. Still, she adds, “there’s a kernel of something good here.”
Dactyl’s real innovation, which isn’t evident from the videos, involves how it transfers learning from simulation to the real world.
OpenAI’s system consists of a humanoid hand, from UK-based Shadow Robot Company, connected to a powerful computer system and an array of cameras and other sensors. Dactyl figures out how to manipulate something using reinforcement learning, which trains a neural network to control the hand based on extensive experimentation.
The latest on artificial intelligence, from machine learning to computer vision and more
Reinforcement learning has produced other impressive AI demos. Most famously, DeepMind, an Alphabet subsidiary, used reinforcement learning to train a program called AlphaGo to play the devilishly difficult and subtle board game Go better than the best human players.
The technique has been used with robots as well. In 2008, Andrew Ng, an AI expert who would go on to hold prominent roles at Google and Baidu, used the technique to make drones perform aerobatics. A few years later, one of Ng’s students, Pieter Abbeel, showed that the approach can teach a robot to fold towels, although this never proved commercially viable. (Abbeel also previously worked part time at OpenAI and still serves as an adviser to the company).
Last year, OpenAI showed Dactyl simply rotating a cube in its hand using a motion learned through reinforcement learning. To wrangle the Rubik’s Cube, however, Dactyl didn’t rely entirely on reinforcement learning. It got help from a more conventional algorithm to determine how to solve the puzzle. What’s more, although Dactyl is equipped with several cameras, it cannot see every side of the cube. So it required a special cube equipped with sensors to understand how the squares are oriented.
Success in applying reinforcement learning to robotics have been hard won because the process is prone to failure. In the real world, it’s not practical for a robot to spend years practicing a task, so training is often done in simulation. But it’s often difficult to translate what works in simulation to more complex conditions, where the slightest bit of friction or noise in a robot’s joints can throw things off.
This is where Dactyl’s real innovation comes in. The researchers devised a more effective way to simulate the complexity of the real world by adding noise, or perturbations to their simulation. In the latest work, this entails gradually adding noise so that the system learns to be more robust to real-world complexity. In practice, it means the robot is able to learn, and transfer from simulation to reality, more complex tasks than previously demonstrated.
Goldberg, the Berkeley professor, who was briefed on the work before it was released, says the simulated learning approach is clever and widely applicable. He plans to try using it himself, in fact.
But he says the limits of the system could have been presented more clearly. The failure rate, for example, was buried deep in the paper, and the video does not show the robot dropping the cube. “But they’re a company, and that’s the difference between academia and companies,” he adds.
Marcus sees Dactyl as the latest in a long line of attention-grabbing AI stunts. He points to a previous announcement from OpenAI, about an algorithm for generating text that was deemed “too dangerous to release,” as evidence the company is prone to oversell its work. “This is not the first time OpenAI has done this,” he says.
OpenAI did not respond to a request for comment.
The surest evidence of how far robots have to go before mastering humanlike dexterity is the small range of repetitive tasks to which robots are limited in industry. Tesla, for example, has struggled to introduce more automation in its plants, and Foxconn has been unable to have robots do much of the fiddly work involved in manufacturing iPhones and other gadgets.
Rodney Brooks, a pioneering figure in robotics and AI who led Rethink Robotics, a now defunct company that tried to make a smarter, easier-to-use manufacturing robot, says academic work involving reinforcement learning is still a long way from being commercially useful.
Brooks, who is now working with Marcus at a robotics startup called Robust.ai, adds that it is easy to misinterpret the capabilities of AI systems. “People see a human doing something, and they know how they can generalize. They see a robot doing something, and they over-generalize,” he says.
“Besides,” Brooks adds. “if [human dexterity] were so close, I would be fucking rich.”
More Great WIRED Stories
- The female founders disrupting the vagina economy
- Silicon Valley cynicism in the age of Trump and Zuckerberg
- Movie talk and the rise of review culture
- Welcome to the “Airbnb for everything” age
- The tech helping dogs learn to “talk” with humans
- ? Prepare for the deepfake era of video; plus, check out the latest news on AI
- ? Upgrade your work game with our Gear team’s fa
Slack calls are having ‘connectivity issues’
Slack has confirmed that “Slack Calls are experiencing some connectivity difficulties right now.” The company said it is working to resolve the issue “as quickly as possible.” The difficulties coincide with the push from tech companies to move workers to remote-only meetings and conference calls, amid the outbreak of COVID-19. Slack did not comment on…
Slack has confirmed that “Slack Calls are experiencing some connectivity difficulties right now.” The company said it is working to resolve the issue “as quickly as possible.” The difficulties coincide with the push from tech companies to move workers to remote-only meetings and conference calls, amid the outbreak of COVID-19.
Slack did not comment on any correlation between the two, or identify what factors are behind the connectivity issues.
Hey @SlackHQ @SlackStatus there’s definitely something up with calls. Everyone keeps getting randomly disconnected, but not all the way. Screen goes black and it just starts randomly playing connect and disconnect sounds.
— matt.js (@mattisadev) March 11, 2020
In a previous blog post outlining Slack’s response to COVID-19, it said “our system architecture is designed to automatically accommodate the surges of traffic throughout the day that this brings to our systems.” The company said its server capacity can handle the demands, as well as the various regions from which users may be logging in. Slack also outlined how the shift to remote may not add a crazy load to its systems.
“The demands on our infrastructure do not change when employees shift away from working together in the same office; there is no difference in load on our systems whether people are connecting from their office, a cellular network, or their homes.”
It added that employees already use an average of nine hours per day, so the volume remains the same.
Establishing an emergency relief fund, Amazon commits to two-week pay for workers affected by COVID-19
Amazon has instituted a new policy which will see all Amazon employees diagnosed with COVID-19 or placed into quarantine receiving up to two-weeks of pay. The additional pay is to “ensure employees have the time they need to return to good health without the worry of lost pay,” the company said in a statement. That…
The additional pay is to “ensure employees have the time they need to return to good health without the worry of lost pay,” the company said in a statement.
That pay is in addition to unlimited paid time off for all hourly employees through the end of March, which the company announced as a policy to its workers last week.
The company also said it was setting up a relief fund with a $25 million contribution to support delivery service partners and drivers along with Amazon Flex participants and seasonal employees.
“We will be offering all of these groups the ability to apply for grants approximately equal to up to two weeks of pay if diagnosed with COVID-19 or placed into quarantine by the government or Amazon,” the company said.
The fund will also support employees and contractors who face financial hardships due to natural disasters, federal emergencies or personal hardship, the company said.
Amazon affiliated workers can apply to receive grant funding ranging from $400 to $5,000 per person.
With this initiative Amazon builds on the commitments it has made as one of several tech companies helping to financially support individuals impacted by the outbreak.
Uber, Salesforce, Cisco, Microsoft, Lyft, Square, Twitter, Facebook, Google, and Apple, have all made commitments to pay hourly and other contingent workers impacted the COVID-19 outbreak. Yesterday, Google announced that it had set up a COVID-19 fund as well.
“As we’re in a transition period in the U.S.—and to cover any gaps elsewhere in the world—Google is establishing a COVID-19 fund that will enable all our temporary staff and vendors, globally, to take paid sick leave if they have potential symptoms of COVID-19, or can’t come into work because they’re quarantined,” writes Adrienne Crowther, Google’s director of workplace services.
“Working with our partners, this fund will mean that members of our extended workforce will be compensated for their normal working hours if they can’t come into work for these reasons. We are carefully monitoring the situation and will continue to assess any adjustments needed over the coming months.”
In addition, Microsoft, Amazon and other Seattle-area companies are partnering with nonprofits and governments to launch a relief fund in response to the outbreak. Amazon and Microsoft committed $1 million apiece to this fund. Microsoft said it would also match employee donations to causes aiding in response to COVID-19.
Superpeer raises $2M to help influencers and experts make money with one-on-one video calls
Superpeer is giving YouTube creators and other experts a new way to make money. The startup announced today that it has raised $2 million in pre-seed funding led by Eniac Ventures, with participation from angel investors including Steven Schlafman, Ankur Nagpal, Julia Lipton, Patrick Finnegan, Justin De Guzman, Chris Lu, Paul Yacoubian and Cheryl Sew…
Superpeer is giving YouTube creators and other experts a new way to make money.
The startup announced today that it has raised $2 million in pre-seed funding led by Eniac Ventures, with participation from angel investors including Steven Schlafman, Ankur Nagpal, Julia Lipton, Patrick Finnegan, Justin De Guzman, Chris Lu, Paul Yacoubian and Cheryl Sew Hoy. It also launched on ProductHunt.
The idea is that if you’re watching a video to learn how to paint, or how to code, or about whatever the topic might be, there’s a good chance you have follow-up questions — maybe a lot of them. Ditto if you follow someone on Twitter, or read their blog posts, to learn more about a specific subject.
Now you could try to submit a question or two via tweet or comment section, but you’re probably not going to get any in-depth interaction — and that’s if they respond. You could also try to schedule a “Can I pick your brain?”-type coffee meeting, but again, the odds aren’t in your favor, particularly when it comes to picking the brain of someone famous or highly in-demand.
With Superpeer, experts who are interested in sharing their knowledge can do so via remote, one-on-one video calls. They upload an intro video, the times that they want to be available for calls and how much they want to charge for their time. Then Superpeer handles the appointments (integrating directly with the expert’s calendar), the calls and the payments, adding a 15% fee on top.
So a YouTube creator could start adding a message at the end of their videos directing fans who want to learn more to their Superpeer page. And if you’re a founder who wants to talk to an experienced designer, executive coach, product manager, marketing/sales expert, VC or other founder, you could start with this list.
Of course, there might be some wariness on both sides, whether you’re an expert who doesn’t want to get stuck on the phone with someone creepy or annoying, or someone who doesn’t want to pay for a call that turns out to be a complete waste of time.
To address this, co-founder and CEO Devrim Yasar (who previously founded collaborative programming startup Koding) said the company has created a user rating system, as well as a way to ask for a refund if you feel that a call violated the terms of service — the calls will be recorded and stored for 48 hours for this purpose.
Superpeer launched in private beta two weeks ago, and Yasar said the startup already has more than 100 Superpeers signed up.