Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI

Deep neural networks (DNNs) can be taught nearly anything, including how to beat us at our own games. The problem is that training AI systems ties up big-ticket supercomputers or data centers for days at a time. Scientists from IBM’s T.J. Watson Research Center think they can cut the horsepower and learning times drastically using “resistive processing units,” theoretical chips that combine CPU and non-volatile memory. Those could accelerate data speeds exponentially, resulting in systems that can do tasks like “natural speech recognition and translation between all world languages,” according to the team.

So why does it take so much computing power and time to teach AI? The problem is that modern neural networks like Google’s DeepMind or IBM Watson must perform billions of tasks in in parallel. That requires numerous CPU memory calls, which quickly adds up over billions of cycles. The researchers debated using new storage tech like resistive RAM that can permanently store data with DRAM-like speeds. However, they eventually came up with the idea for a new type of chip called a resistive processing unit (RPU) that puts large amounts of resistive RAM directly onto a CPU.

Read more

Annual sales of drones in the U.S. will hit 2.5 million this year and swell to 7 million by 2020, according to a projection from the Federal Aviation Administration.

Unmanned aircraft purchases are growing both for hobbyists and for commercial ventures that perform inspections, assist farmers, and survey construction sites, according to the agency’s annual forecast of aviation activity, released on Thursday.

Read more

Companies that sell personal data should pay a percentage of the resulting revenue into a Data Mining Royalty Fund that would provide annual payments to U.S. citizens, much as the Alaska Permanent Fund distributes oil revenues to Alaskans.

A viral video released in February showed Boston Dynamics’ new bipedal robot, Atlas, performing human-like tasks: opening doors, tromping about in the snow, lifting and stacking boxes. Tech geeks cheered and Silicon Valley investors salivated at the potential end to human manual labor.

Shortly thereafter, White House economists released a forecast that calculated more precisely whom Atlas and other forms of automation are going to put out of work. Most occupations that pay less than $20 an hour are likely to be, in the words of the report, “automated into obsolescence.”

Read more

They are pushing for fully automated robot Cargo ships. Now, we have robots to load and unload cargo ships. In a few years there probably wont be a single person left working on a dock.

TraPac LLC’s Los Angeles shipping terminal offers a window to how coming global trade will move: using highly automated systems and machinery to handle a flood of goods amid new free-trade accords.

Read more

(Photo credit: AP Photo/Jack Plunkett, File)

Astro Teller is tired of the paranoia surrounding artificial intelligence and robotics. The famous computer scientist’s sensitivity around the topic may be understandable considering he bears the brunt of some of that skepticism as the head of X, the Alphabet (and formerly Google) moonshot factory working on many of the company’s futuristic AI and robotics projects.

This past weekend, Teller, whose official title is “captain of moonshots,” took to the stage at the inaugural Silicon Valley Comic Con hoping to dispel some of these misconceptions around AI. His physician wife, Danielle Teller, presented alongside him on some of the fear mongering associated with genetic engineering in humans. After their presentation, the Tellers sat down with FORBES to go deeper on the issue to explain what they hoped to accomplish with their talk.

Read more

A novel written by artificial intelligence was a finalist in Japan’s Hoshi Shinichi Literary Award. The award is named after Hoshi Shinichi, a Japanese science fiction author whose books include The Whimsical Robot and Greetings from Outer Space. The unique contest accepts submissions from humans and machines, and judges for the prize, now in its third year, weren’t told which novels were written by humans and which were penned by human-AI teams. This year was the first time the committee received submissions written by AI programs.

The AI novel is called The Day A Computer Writes A Novel, or Konpyuta ga shosetsu wo kaku hi in Japanese. It was co-written by Hishoshi Matsubara, a professor of computer science, along with his team at Future University Hakodate in Japan. According to the LA Times, their AI wrote four books, of which one made it past the first round of the prize. It was one of 1450 submissions, 11 of which were written with the help of AI programs.

According to reports, 80% of the novel had human involvement, as Matsubara and his team did the research for the novel, decided on the plot and developed the characters. The novel’s text was written entirely by the AI. The Professor’s team entered words and phrases from a sample novel into a computer in order for the AI to construct a new novel similar to it, Slate reports.

Read more