Shedding drivers would save Uber a lot of money. Ride-hailing service Uber has sounded out car companies about placing a large order for self-driving cars, an auto industry source has said. "They wanted autonomous cars," the source, who declined to be named, said. "It seemed like they were shopping around." Loss-making Uber would make drastic savings on its biggest cost -- drivers -- if it were able to incorporate self-driving cars into its fleet.

We analyse the matrix factorization problem. Given a noisy measurement of a product of two matrices, the problem is to estimate back the original matrices. It arises in many applications such as dictionary learning, blind matrix calibration, sparse principal component analysis, blind source separation, low rank matrix completion, robust principal component analysis or factor analysis. It is also important in machine learning: unsupervised representation learning can often be studied through matrix factorization. We use the tools of statistical mechanics - the cavity and replica methods - to analyze the achievability and computational tractability of the inference problems in the setting of Bayes-optimal inference, which amounts to assuming that the two matrices have random independent elements generated from some known distribution, and this information is available to the inference algorithm. In this setting, we compute the minimal mean-squared-error achievable in principle in any computational time, and the error that can be achieved by an efficient approximate message passing algorithm. The computation is based on the asymptotic state-evolution analysis of the algorithm. The performance that our analysis predicts, both in terms of the achieved mean-squared-error, and in terms of sample complexity, is extremely promising and motivating for a further development of the algorithm.

Inbox by Gmail has a few useful features but probably the most interesting is Smart Reply, which was expanded to the desktop this week. As impressive as the A.I.-powered suggestions for replies to your emails are, they're just not very useful yet. Sure, Smart Replies tend to offer relevant options, but they're not very useful in most real-world cases. Open an invitation to party and you might get'Sounds good!,' 'Can't wait,' and Sorry, I can't be there!' as the suggestions. By the time I've added'Hi Bob,' and a few other bits of personalization, I might as well have written the whole thing myself.

The number of new malware variations that pop up each day runs somewhere between 390,000 (according to AV-TEST Institute) and one million (according to Symantec Corporation). These are new strains of malware that have not been seen in the wild before. Even if we consider just the low end figure, the situation is still dire. Google Now is about to get a lot better in the future, aiming to serve Android users even when they're offline. With smartphones increasingly gaining ground, digital assistants have become widely popular and heavyweight companies are competing to deliver the best software in this category.

For outsiders, the idea of artificial intelligence in the court room sounds horrifying. AI, however, has been pushing its way into law for decades. The next time you apply to claim child benefits, you may be met with the unexpected: robots. Most might not even notice it, but the future of law is heavily tied to Artificial Intelligence. In fact, the slow integration of AI into the legal sphere has been happening for decades, and several magazines, news sources and committees have been built around the topic.

At San Francisco's first fully automated restaurant, meals appear in little glass cubbies, just 90 seconds after customers order and pay on wall-mounted iPads. It's a human-less experience – no waitstaff, no cashier, no one to get your order wrong and no one to tip. The moment before the meal appears, the see-through display screen that fronts the cubbies goes black for the few seconds when you might catch sight of the hand that feeds you. Eatsa has not yet achieved total automation. The company admits it employs a small kitchen staff, and one employee is present in the front of the house, answering questions about how to order and dodging questions about what's going on behind the wall of magic cubbies.

Unexpected convergent consequences…this is what happens when eight different exponential technologies all explode onto the scene at once. This post (sixth in a series of seven) is a look at robotics. Be sure to read the first five posts if you haven't already: When the World Is Wired: The Magic of the Internet of Everything Where Artificial Intelligence Is Now and What's Just Around the Corner The Near Future of VR and AR: What You Need to Know Drones Have Reached at Tipping Point--Here's What Happens Next How 3D Printing Is Transforming the Way We Make Things An expert might be reasonably good at predicting the growth of a single exponential technology (e.g., 3D Printing), but try to predict the future when AI, robotics, VR, drones, and computation are all doubling, morphing and recombining…You have a very exciting (read: unpredictable) future. This post is the result of an interview with Rodney Brooks on the top five recent robotics breakthroughs (2012-2015) and the top five anticipated robotics breakthroughs (2016-2018). Rodney is the Panasonic Professor of Robotics at MIT.

Rapid advances in artificial intelligence technology are raising ethical questions, as pointed out by Dr Jason Millar ("The momentous advance in artificial intelligence demands a new set of ethics", Comment). He asks whether it is desirable to develop autonomous systems that operate beyond human control. Other ethical dilemmas may arise sooner than we think. While many have poured scorn on the idea that robots could possess consciousness, if consciousness can be interpreted as an algorithm – a series of logical cause-and-effect statements – then, because the output of an algorithm is platform-independent, there is no reason in principle why that algorithm should not operate in a robot. There is a debate as to whether brain activity is algorithmic, but other forms of biological information processing are and there is no convincing evidence to the contrary.

A report shown at the 2016 World Economic Forum in January says millions of jobs will be lost to robots in the next few years. When thinking about who is most vulnerable, factory workers, drivers, and pilots come to mind. Surely the jobs requiring a human touch, such as artists, entertainers, and managers, will stick around, right? Maybe some of those jobs will be safe. Managers, not so much; very soon, robots will be replacing humans in top management positions, even up to the CEO level.

Updated 15 March 2016: Today concludes the five'Go' matches played by AlphaGo, an AI system built by DeepMind and South Korean champion, Lee Sedol. AlphaGo managed to win the series of games 4-1. 'Go' is a strategy-led board game in which two players aim to gather and surround the most territory on the board. The game is said to require a certain level of intuition and be considerably more complex than Chess. The first three games were won by AlphaGo with Sedol winning the fourth round, but still unable to claim back a victory.