AI2-THOR Interactive Simulation Teaches AI About Real World

AI2-THOR, an interactive simulation based on home environments, can prepare AI for real-world challenges Image: Roozbeh Mottaghi, Eric Kolve / Allen Institute for Artificial Intelligence Training a robot butler to make the perfect omlette could require breaking a lot of eggs and throwing out many imperfect attempts in a real-life kitchen. That’s why researchers have been rolling out virtual training grounds as a more efficient alternative to putting AI agents through costly and time-consuming experiments in the real world. Virtual environments could prove especially useful in training the most popular AI based on machine learning algorithms that often require thousands of trial-and-error runs to learn new skills. Companies such as Waymo have already built their own internal simulators with virtual roads and traffic intersections to train their AI to safely take the wheel of self-driving cars. But a new, open-source virtual training ground called AI2-THOR enables AI agents to learn how to interact with objects in familiar home Continue reading AI2-THOR Interactive Simulation Teaches AI About Real World

Why Ethical Robots Might Not Be Such a Good Idea After All

The risks that a robot’s ethics might be compromised by unscrupulous actors are so great as to raise serious doubts over the wisdom of embedding ethical decision making in real-world safety critical robots Photo: Bristol Robotics Lab This is a guest post. The views expressed here are solely those of the author and do not represent positions of IEEE Spectrum or the IEEE. This week my colleague Dieter Vanderelst presented our paper: “The Dark Side of Ethical Robots” at AIES 2018 in New Orleans. I blogged about Dieter’s very elegant experiment here, but let me summarize. With two NAO robots he set up a demonstration of an ethical robot helping another robot acting as a proxy human, then showed that with a very simple alteration of the ethical robot’s logic it is transformed into a distinctly unethical robot—behaving either competitively or aggressively toward the proxy human. Here are our paper’s key conclusions: The ease of transformation Continue reading Why Ethical Robots Might Not Be Such a Good Idea After All

Robots Ready to Ski, Paint, and Clean at South Korea’s 2018 Winter Olympics

South Korea’s best roboticists have spent two years preparing robots for a wide range of roles at the 2018 Winter Olympics Photo: Ang Young-suk/Yonhap/AP Photo: Ang Young-suk/Yonhap/AP The Handoff: The robot DRC-Hubo+ receives the Olympic flame from UCLA professor Dennis Hong. Spectators who getlost in Olympic Plaza in Pyeongchang next month can ask for directions from a nearby guide that speaks four languages. Thirsty patrons who visit Gangneung Media Village can order drinks for delivery. And they can do all of this without talking to another human. South Korea is going big on robotics for the 2018 Winter Olympics, which begin on 9 February. Organizers will deploy about 80 robots at the games to showcase the nation’s leader­ship in advanced robotics research. Eight companies—with US $1.5 million in sponsorship from the South Korean government—have been working on projects for the games since 2016. The roboticists who built all of these Continue reading Robots Ready to Ski, Paint, and Clean at South Korea’s 2018 Winter Olympics

Why You Should Fear ‘Slaughterbots’—A Response

Lethal autonomous weapons are not science fiction; they are a real threat to human security that we must stop now Image: Slaughterbots/YouTubeA scene from “Slaughterbots,” a video produced by the Future of Life Institute to illustrate the dangers of autonomous weapons, depicts an explosive-carrying micro-drone that uses AI to autonomously target specific individuals. This is a guest post. The views expressed here are solely those of the authors and do not represent positions of IEEE Spectrum or the IEEE. Paul Scharre’s recent article “Why You Shouldn’t Fear ‘Slaughterbots’” dismisses a video produced by the Future of Life Institute, with which we are affiliated, as a “piece of propaganda.” Scharre is an expert in military affairs and an important contributor to discussions on autonomous weapons. In this case, however, we respectfully disagree with his opinions. Why we made the video We have been working on the autonomous weapons issue for several years. We Continue reading Why You Should Fear ‘Slaughterbots’—A Response

This AI Hunts Poachers

The elephant’s new protector is PAWS, a machine-⁠learning and game-theory system that predicts where poachers are likely to strike Illustration: MCKIBILLO Illustration: MCKIBILLO Every year, poachers kill about 27,000 African elephants—an astounding 8 percent of the population. If current trends continue, these magnificent animals could be gone within a decade. The solution, of course, is to stop poachers before they strike, but how to do that has long confounded authorities. In protected areas like wildlife preserves, elephants and other endangered animals may roam far and wide, while rangers can patrol only a small area at any time. “It’s a two-part problem,” explains Milind Tambe, a computer scientist at the University of Southern California, in Los Angeles. “Can you predict where poaching will happen? And can you [target] your patrols so that they’re unpredictable, so that the poachers don’t know the rangers are coming?” To solve both parts of the problem, Tambe and Continue reading This AI Hunts Poachers

Hacked Dog Pics Can Play Tricks on Computer Vision AI

An MIT student lab shows how to trick computer vision AI so it sees the wrong objects in pictures Photo: Dageldog/iStock Tricking Google’s computer vision AI into seeing a dog as a pair of human skiers may seem mostly harmless. But the possibilities become more unnerving when considering how hackers could trick a self-driving car’s AI into seeing a plastic bag instead of a child up ahead. Or making future surveillance systems overlook a gun because they see it as a toy doll. An independent AI research group run by MIT students has demonstrated a new way to fool the computer vision algorithms that enable AI systems to see the world—an approach that could prove up to 1000 times as fast as other existing ways of hacking “black box” systems whose inner workings remain hidden to outsiders. That idea of a black box perfectly describes the neural networks behind the deep learning algorithms enabling computer vision services for Continue reading Hacked Dog Pics Can Play Tricks on Computer Vision AI

Artificial Intelligence Predicts Outcomes of Chemical Reactions

An AI built by IBM could help organic chemists find new ways to synthesize drugs Photo: IBM Research By thinking of atoms as letters and molecules as words, artificial intelligence software from IBM is now employing the same methods computers use to translate languages to predict outcomes of organic chemical reactions, which could speed the development of new drugs. In the past 50 years, scientists have tried to teach computers how chemistry works so that computers can help predict the results of organic chemical reactions. However, organic chemicals can be extraordinarily complex, and simulations of their behavior can prove time-consuming and inaccurate. Instead, researchers at IBM took the kind of AI program normally used to translate languages and applied it toward organic chemistry. “Instead of translating English into German or Chinese, we had the same artificial intelligence technology look at hundreds of thousands or millions of chemical reactions and had Continue reading Artificial Intelligence Predicts Outcomes of Chemical Reactions

DARPA Seeking AI That Learns All the Time

The agency wants ideas for turning computers into lifelong learners Illustration: iStockphoto Earlier this month a self-driving shuttle in Las Vegas patiently waited as a delivery truck backed up, then backed up some more, then backed right into it. Inconveniently for the roboshuttle’s developer Navya, this happened within hours of the shuttle’s inauguration ceremony. The real problem is that the shuttle can’t learn from the incident the way a human would: immediately and without forgetting how to do everything else in the process. The U.S. Defense Advanced Research Projects Agency (DARPA) is looking to change the way AI works through a program it calls L2M, or Lifelong Learning Machines. The agency is looking for systems that learn continuously, adapt to new tasks, and know what to learn and when. “We want the rigor of automation with the flexibility of the human,” says the program’s director Hava T. Siegelmann. The US Continue reading DARPA Seeking AI That Learns All the Time

AI-Powered Microscope Counts Malaria Parasites in Blood Samples

Silicon Valley teams up with a Chinese microscope manufacturer to deploy deep learning to diagnose malaria Photo: Andrew H. Kim/Intellectual VenturesRoxanne Rees-Channer, a research biochemist, inserts a cassette into the EasyScan GO at the Hospital for Tropical Diseases in London, where the AI-powered microscope is being tested. Today, a Chinese manufacturer and a venture backed by the Bill & Melinda Gates Foundation will announce plans to commercialize a microscope that uses deep learning algorithms to automatically identify and count malaria parasites in a blood smear within 20 minutes. AI-powered microscopes could speed up diagnosis and standardize detection of malaria at a time when the mosquito-borne disease kills almost half a million people per year. An experimental version of the AI-powered microscope has already shown that it can detect malaria parasites well enough to meet the highest World Health Organization microscopy standard, known as competence level 1. That rating means that it performs on par Continue reading AI-Powered Microscope Counts Malaria Parasites in Blood Samples

AI Startup Embodied Intelligence Wants Robots to Learn From Humans in Virtual Reality

UC Berkeley spinoff wants to use AI and VR to teach robots new skills Photo: Elena Zhukova/Embodied IntelligenceEmbodied Intelligence wants to use AI and VR to teach robots new skills, like how to manipulate wires, much faster. Depending on who you ask, robotic grasping has been solved for a while now. That is, the act of physically grasping an object, not dropping it, and then doing something useful is a thing that robots are comfortable with. The difficult part is deciding what to grasp and how to grasp it, and that can be very, very difficult, especially outside of a structured environment. This is a defining problem for robotics right now: Robots can do anything you want, as long as you tell them exactly what that is, every single time. In a factory where robots are doing the exact same thing over and over again, this isn’t so much of Continue reading AI Startup Embodied Intelligence Wants Robots to Learn From Humans in Virtual Reality