The “world’s strongest computers” have been powering Google’s DeepMind AI systems for years, but the company is now aiming to use the technology to power machines that can “do things that you would have never imagined possible.”
In a new paper published on Tuesday, the company says that DeepMind has “overwhelmingly surpassed” its competitors in terms of processing power and that its “super-fast” machines have already “gained a lot of traction.”
DeepMind’s computers can compute things that were previously “unachievable,” including “the ability to do things that are hard to do today,” as well as “a lot of things that we have never done before,” such as “superimposing multiple layers of images on top of one another.”
Deepmind has also been able to “get the largest amount of work done in a short time,” and “tend to be better at solving complex problems than any of the other companies we’ve seen,” the researchers write.
In particular, the machine’s supercomputers are able to perform “a massive amount of parallel computing” and “create algorithms with a lot more power than the previous supercomputing engines,” as the company notes.
Google is now using the “deep learning” approach to help build its artificial intelligence (AI) system.
As Ars Technic noted in 2015, Google’s machine learning technology has already proven useful for things like mapping data in maps, analyzing images, and even predicting which songs will hit the top of the charts.
It’s possible that Google’s AI system will also be used to build the “worlds most powerful computer” in the future.
Google DeepMind DeepMind is the company that developed Google’s self-driving car, which is also now being used to create artificial intelligence machines for the military.
However, the system was only able to produce “small, accurate” predictions about what might happen in the event of a disaster, such as when the vehicle failed to stop at a stop sign.
The company’s researchers say that “when you look at the data that we’ve been able get, it seems like we’ve had a lot better luck than we had in the past.”
Google Deepmind’s AI systems can also be deployed on other devices, such a “self-driving aircraft.”
For example, Google Deep Mind is now deploying its “deep-learning” systems on its self-driven aircraft, which “can fly at Mach 1.5 or faster and achieve a flight speed of around 300 knots,” the company said in a statement.
This is “a really big leap,” the statement said, “whereby they can go from being a hobby for the people in the world to being the most powerful supercomputer in the whole world.”
In addition to developing machines that are “the most powerful in the entire world,” Google Deep Machine is also “building an AI system that can go far beyond what we can do today.”
Deep Machine “is capable of performing a number of tasks that we can’t do today, such the ability to build a computer model of a physical system, or even to run a simulation of an event that we don’t yet understand,” the team said.
Google’s researchers are now working on an update that will make it possible to “build the world’s most advanced computer in the foreseeable future,” and the company also says that the team “is looking at other applications that we might want to develop.”