At Google, robots go to school and learn using artificial intelligence algorithms

At Google, robots go to school and learn using artificial intelligence algorithms

MOUNTAIN VIEW, CA – Researchers here in a Google lab recently ordered a robot that made a burger out of various plastic toy components.

The mechanical arm knew enough to add ketchup after the meat and before the lettuce, but he thought the correct way to do it was to put the entire bottle inside the burger.

While this robot won’t be working as a cook anytime soon, it’s an even bigger achievement that Google engineers announced on Tuesday. Using recently developed artificial intelligence software known as Big Language Models, researchers say they have been able to design robots that can help humans with a wider range of everyday tasks.

Instead of presenting a list of washing instructions — directing all of the robot’s movements one by one — robots can now respond to complete requests, like a human.

At a demonstration last week, a researcher told a robot, “I’m hungry, can you bring me a snack?” The robot then proceeded to search a cafeteria, open a drawer, find a bag of chips, and bring it to the human.

Google executives and researchers say it’s the first time language models have been integrated into bots.

“This is a fundamentally different model,” says Brian Eichter, a research scientist at Google and one of the authors of a new research paper released Tuesday describing the company’s progress.

Robots are already common. Millions of them work in factories around the world, but they follow specific instructions and usually focus only on one or two tasks, such as moving a product down an assembly line or welding two pieces of metal together. The race to build a robot that can do a range of everyday tasks, and to learn on the job, is much more complex. Tech companies big and small have been building such general-purpose robots for years.

Big Tech builds AI with bad data. So scientists sought better data.

Linguistic models work by taking huge amounts of text uploaded to the Internet and using it to train AI programs to guess the types of responses that might come after certain questions or comments. Models have become so good at predicting the correct response that dealing with one is often like having a conversation with a knowledgeable human being. Google and other companies, including OpenAI and Microsoft, have poured resources into building better models and training them on larger sets of text in multiple languages.

The work is controversial. In July, Google fired one of its employees who claimed that he believed the program was conscious. The consensus among AI experts is that the models are not sensitive, but many are concerned that they show bias because they have been trained in massive amounts of human-generated unfiltered text.

Some language models have been shown to be racist or sexist, or are easily manipulated to deliver hate speech or lies when prompted for correct statements or questions.

Linguistic models in general can give bots knowledge of high-level planning steps, said Deepak Pathak, an assistant professor at Carnegie Mellon University who studies artificial intelligence and robotics and has been commenting on the field, not Google specifically. But these models will not give the robots all the information they need – for example, how much force to apply when opening the refrigerator. That knowledge has to come from somewhere else.

“It only solves the question of high-level planning,” he said.

However, Google is going ahead, and has now combined language models with some of its bots. Now, instead of having to code specific technical instructions for each task the robot can perform, researchers can talk to them in everyday language. More importantly, the new software helps bots analyze complex, multi-step instructions on their own. Now, robots can interpret instructions they’ve never heard before and come up with logical responses and reactions.

These robots are trained in artificial intelligence. They became racist and sexist.

Robots that can use language models could change how manufacturing and distribution facilities are managed, said Zack Stewart Rogers, assistant professor of supply chain management from Colorado State University.

“Human and robot working together is always the most productive” now, he said. Robots can do manual lifting. Humans can do accurate troubleshooting.”

If robots are able to figure out complex tasks, that could mean distribution centers could be smaller, with fewer humans and more robots. That could mean fewer jobs for people, although Rogers points out that generally when there is a contraction due to automation in one area, jobs are created in others.

There may be a long way to go. Artificial intelligence techniques such as neural networks and reinforcement learning have been used to train robots for years. It has led to some breakthroughs, but progress is still slow. Google bots aren’t anywhere ready for the real world, and in interviews, Google researchers and executives have said time and time again that they simply run a research lab and have no plans to commercialize the technology yet.

But it’s clear that Google and other big tech companies have a serious interest in robotics. Amazon is using several robots in its warehouses, is experimenting with delivery of drones, and earlier this month agreed to buy the Roomba vacuum cleaner robot for $1.7 billion. (Amazon founder Jeff Bezos owns The Washington Post.)

Tesla says it’s building a ‘friendly’ robot that will do menial tasks, and it won’t resist

Tesla, which has developed some self-driving features for its cars, is also working on general-purpose robots.

In 2013, Google launched into a massive spending spree, buying several robotics companies, including Boston Dynamics, the maker of robotic dogs that often go viral on social media. But the executive director in charge of the program was accused of sexual misconduct, and left the company soon after. In 2017, Google sold Boston Dynamics to Japanese telecommunications and technology investment giant Softbank. The hype about smarter robots built by the most powerful tech companies has faded.

On the Language Model project, Google researchers worked alongside who everyday robots, a separate but wholly owned company within Google that builds bots that can perform a range of “repetitive” and “tedious” tasks. Bots are already at work in many Google cafes, scanning counters and dumping trash in the trash.


Google researchers worked alongside researchers at Everyday Robots, a separate but wholly owned company within Google. An earlier version of this story incorrectly named the company. This story has been updated.

#Google #robots #school #learn #artificial #intelligence #algorithms

Leave a Comment

Your email address will not be published. Required fields are marked *