Nasa has cancelled the first ever all-female spacewalk, saying it doesn't have enough spacesuits ready in the right size. Two of the astronauts currently on the International Space Station – Christina Koch and Anne McClain – had been set to head out of the floating lab to instal powerful new batteries on the outside to charge up from its solar arrays. But now the space agency says a problem with spacesuits will mean Ms McClain is unable to take part in the spacewalk, and she will be replaced by a man, fellow Nasa astronaut Nick Hague. We'll tell you what's true. You can form your own view.

AI fuzzing uses machine learning and similar techniques to find vulnerabilities in an application or system. Fuzzing has been around for a while, but it's been too hard to do and hasn't gained much traction with enterprises. Adding AI promises to make the tools easier to use and more flexible. The good news is that enterprises and software vendors will have an easier time finding potentially exploitable vulnerabilities in their systems so they can fix them before bad guys get to them. The bad news is that the bad guys will have access to this technology as well and will soon start to find zero-day vulnerabilities on a massive scale.

Dozens of scientists, health care professionals and academics have written a letter to the U.N. calling for an international ban of autonomous killer robots, saying recent advances in artificial intelligence "have brought us to the brink of a new arms race in lethal autonomous weapons." The letter, which has been signed by more than 70 health care professionals and was put together by the Future of Life Institute, states that lethal autonomous weapons could fall into the hands of terrorists and despots, lower the barrier to armed conflict and "become weapons of mass destruction enabling very few to kill very many." "Furthermore, autonomous weapons are morally abhorrent, as we should never cede the decision to take a human life to algorithms," the letter continues. "As healthcare professionals, we believe that breakthroughs in science have tremendous potential to benefit society and should not be used to automate harm. We therefore call for an international ban on lethal autonomous weapons."

In this thesis we discuss machine learning methods performing automated variable selection for learning sparse predictive models. There are multiple reasons for promoting sparsity in the predictive models. By relying on a limited set of input variables the models naturally counteract the overfitting problem ubiquitous in learning from finite sets of training points. Sparse models are cheaper to use for predictions, they usually require lower computational resources and by relying on smaller sets of inputs can possibly reduce costs for data collection and storage. Sparse models can also contribute to better understanding of the investigated phenomenons as they are easier to interpret than full models.

The history of AI is often told as the story of machines getting smarter over time. What's lost is the human element in the narrative, how intelligent machines are designed, trained, and powered by human minds and bodies. In this six-part series, we explore that human history of AI--how innovators, thinkers, workers, and sometimes hucksters have created algorithms that can replicate human thought and behavior (or at least appear to). While it can be exciting to be swept up by the idea of super-intelligent computers that have no need for human input, the true history of smart machines shows that our AI is only as good as we are. On 14 February 1946, journalists gathered at the Moore School of Engineering at the University of Pennsylvania to witness a public demonstration of one of the world's first general-purpose electronic digital computers: the Electronic Numerical Integrator and Computer (ENIAC).

GITAI is a robotics startup with offices in Japan and the United States that's developing tech to put humanoid telepresence robots in space to take over for astronauts. Today, GITAI is announcing a joint research agreement with JAXA (the Japanese Aerospace Exploration Agency) to see what it takes for robots to be useful in orbit, with the goal of substantially reducing the amount of money spent sending food and air up to those demanding humans on the International Space Station. It's also worth noting that GITAI has some new hires, including folks from the famous (and somewhat mysterious) Japanese bipedal robot company SCHAFT. A quick reminder about SCHAFT: The company was founded by members of the JSK Laboratory at the University of Tokyo in order to build a robot to compete in the DARPA Robotics Challenge Trials in 2013. SCHAFT won the DRC Trials by a substantial margin, scoring 27 points out of a possible 32, 7 more points than the second place team (IHMC).

Government usually isn't the place to look for innovation in IT or new technologies like artificial intelligence. But Ott Velsberg might change your mind. As Estonia's chief data officer, the 28-year-old graduate student is overseeing the tiny Baltic nation's push to insert artificial intelligence and machine learning into services provided to its 1.3 million citizens. "We want the government to be as lean as possible," says the wiry, bespectacled Velsberg, an Estonian who is writing his PhD thesis at Sweden's Umeå University on how to use AI in government services. Estonia's government hired Velsberg last August to run a new project to introduce AI into various ministries to streamline services offered to residents.

AI has a data quality problem. In a survey of 179 data scientists, over half identified addressing issues related to data quality as the biggest bottleneck in successful AI projects. Big data is so often improperly formatted, lacking metadata, or "dirty," meaning incomplete, incorrect, or inconsistent, that data scientists typically spend 80 percent of their time on cleaning and preparing data to make it usable, leaving them with just 20 percent of their time to focus on actually using data for analysis. This means organizations developing and using AI must devote huge amounts of resources to ensuring they have sufficient amounts of high-quality data so that their AI tools are not useless. As policymakers pursue national strategies to increase their competitiveness in AI, they should recognize that any country that wants to lead in AI must also lead in data quality.

In medicine, diseases can be detected at a much earlier stage, and we can support the elderly to live a more independent life, simply by identifying deviations from their usual behaviour and body movements. The UK Government recently announced that AI could help the National Health Service predict those in an early stage of cancer, to ultimately prevent thousands of cancer-related deaths by 2033. The algorithms will examine medical records, habits and genetic information pooled from health charities, the NHS and AI. Virtual nurses could transform patient care, being available round the clock to answer questions, monitor patients and provide quick answers. Beyond healthcare, AI could inform a better allocation of resources in energy, logistics and transport, as well as support the digital advertising industry with more efficient marketing.

In autonomous embedded systems, it is often vital to reduce the amount of actions taken in the real world and energy required to learn a policy. Training reinforcement learning agents from high dimensional image representations can be very expensive and time consuming. Autoencoders are deep neural network used to compress high dimensional data such as pixelated images into small latent representations. This compression model is vital to efficiently learn policies, especially when learning on embedded systems. We have implemented this model on the NVIDIA Jetson TX2 embedded GPU, and evaluated the power consumption, throughput, and energy consumption of the autoencoders for various CPU/GPU core combinations, frequencies, and model parameters. Additionally, we have shown the reconstructions generated by the autoencoder to analyze the quality of the generated compressed representation and also the performance of the reinforcement learning agent. Finally, we have presented an assessment of the viability of training these models on embedded systems and their usefulness in developing autonomous policies. Using autoencoders, we were able to achieve 4-5 $\times$ improved performance compared to a baseline RL agent with a convolutional feature extractor, while using less than 2W of power.