A man-made power island in the middle of the North Sea that could supply electricity for 80 million people, a robot that could read your mind and spot you noticing it made a mistake, and a DNA-based computer that grows as it computes? Go figure!
Three European countries are reportedly planning to build an artificial power island in a shallow area of the North Sea called the Dogger Bank. The island would collect electricity from thousands of wind turbines surrounding it. The 6-square-kilometer stone-and-sand structure backed by Denmark’s Energinet.dk and Germany and the Netherland’s TenneT would collect and send to land between 70 to 100 gigawatts in the long term. The lower number represents enough power to supply 80 million people. “This project can significantly contribute to a completely renewable supply of electricity in Northwest Europe,” Mel Kroon, CEO of TenneT, said in a news release.
Top image credit: Energinet.dk.
A team of researchers from MIT’s Computer Science and Artificial Intelligence Laboratory and Boston University are working on a system that could enable humans to communicate with robots with just their thoughts. The first iteration of the technology can accomplish “relatively simple binary-choice activities” by monitoring brain activity with an electroencephalography cap. The machine-learning system can spot brainwaves generated when humans notice a mistake. Users watching a Baxter robot from Rethink Robotics performing a task were able to alert the machine that it had made a mistake just with their minds. “As you watch the robot, all you have to do is mentally agree or disagree with what it is doing,” CSAIL Director Daniela Rus told MIT News. “You don’t have to train yourself to think in a certain way — the machine adapts to you, and not the other way around.” What’s more, “if the robot’s not sure about its decision, it can trigger a human response to get a more accurate answer,” according to Rus’ colleague Stephanie Gil. “These signals can dramatically improve accuracy, creating a continuous dialogue between human and robot in communicating their choices.” Said Rus: “Imagine being able to instantaneously tell a robot to do a certain action, without needing to type a command, push a button or even say a word. A streamlined approach like that would improve our abilities to supervise factory robots, driverless cars, and other technologies we haven’t even invented yet.”
Scientists at The University of Manchester in England say they have found a way to build from DNA molecules “a super-fast form of computer that ‘grows as it computes.’ ” The team writes in the Journal of the Royal Society Interface: “The design exploits DNA’s ability to replicate to execute an exponential number of computational paths in P time.” Professor Ross King from Manchester’s School of Computer Science explained it further: “Imagine a computer is searching a maze and comes to a choice point, one path leading left, the other right,” he said. “Electronic computers need to choose which path to follow first. But our new computer doesn’t need to choose, for it can replicate itself and follow both paths at the same time, thus finding the answer faster.” King said that “as DNA molecules are very small a desktop computer could potentially utilize more processors than all the electronic computers in the world combined — and therefore outperform the world’s current fastest supercomputer, while consuming a tiny fraction of its energy.”
Google’s Deep Mind engineers describe in a new paper an algorithm called PathNet that supports “transfer, continual and multi-task learning.” It could serve as a stepping stone to the emergence of “super neural networks” and general artificial intelligence. “A plausible requirement for artificial general intelligence is that many users will be required to train the same giant neural network on a multitude of tasks” without “catastrophic forgetting,” the authors write. They say that PathNet is “a first step in this direction.” They continue: “This is the most efficient way for the network to gain experience, because such a network can reuse existing knowledge instead of learning from scratch for each task. To achieve this, we propose that each user of the giant net be given a population of agents whose job it is to learn the user-defined task as efficiently as possible. Agents will learn how best to re-use existing parameters in the environment of the neural network by executing actions within the neural network.”
U.S. government researchers have developed a reusable foam called Oleo Sponge that can “easily” soak up oil from water even below the surface. The material’s “unprecedented” possibilities could come in handy during oil spill cleanups. Working at the U.S. Department of Energy’s (DOE) Argonne National Laboratory, the team created the super sponge by coating its internal cavities with a thin layer of a special polymer. The layer “serves as the perfect glue for attaching the oil-loving molecules,” according to an Argonne news release. “The technique offers enormous flexibility, and can be adapted to other types of cleanup besides oil in seawater,” according to Argonne chemist Jeff Elam. “You could attach a different molecule to grab any specific substance you need.”