When Singapore decided that they needed a new smart water assessment network to track pollution in their reservoirs, they obviously went with a robot, because otherwise you wouldn't be reading about it here. They also decided that the robot had to be "aesthetically pleasing" in order to "promote urban livability." The answer came from researchers at the National University of Singapore (NUS), who proposed developing a Smart Water Assessment Network: Yes, that's right, a SWAN. The researchers, from NUS Environmental Research Institute and Tropical Marine Science Institute, had developed and tested a version of the swanbots back in 2016. Now they've decided to deploy them full time across five different reservoirs in Singapore, where water is a particularly precious resource.

There are a lot of things that chatbots have yet to master and high on the list is small talk. But researchers at Facebook think the best way to make software prattle away is to give it a personality. Workers were asked to chat in pairs and to give statements describing themselves, including their likes and dislikes. The crowdworkers' chatter was linked to these description statements and used to train the chatbots.

By default, both iOS and Android let you access their digital assistants--Siri and Google Assistant, respectively--right from the lock screen (unless the phone owner has disabled the feature). To pull up the assistant on an iPhone other than the X, press and hold the Home button (for the iPhone X, press and hold the Power button instead). On Android, tap and hold in the bottom left corner of the display (where the microphone icon apppears) and then drag your finger up to the middle of the screen. Once you've accessed Siri or Google Assistant, try saying "Call mom," "Call home," or another command that might access one of the phone's owner's contacts. Siri also has a trick that Google Assistant doesn't: Ask "Whose phone is this?" to bring up contact details for the owner.

Criminal recidivism is when a released criminal goes back to crime. From charging crimes through probation, the criminal justice system is constantly looking for ways to better predict which criminals are more likely to remain legal on release and who is a risk of recidivism. Bias can create inaccuracies through weighing variables incorrectly, and machine learning might provide a way of limiting bias and improving recidivism predictions. A recent study by Julia Dressel and Hany Farid, published in Science Advances, points to the limitations of deterministic algorithms with fixed parameters for the task of such predictions. The study analyzes the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) software, a package used by court systems to predict the likelihood of recidivism in criminal defendants. The lessons learned lead me to a discussion about the promise of machine learning (ML) systems – specifically, deep learning.

One day after an American drone strike killed a leader of the militant Haqqani network in northwestern Pakistan, United States officials on Thursday rejected a claim by Pakistan that the strike had targeted an Afghan refugee camp. There were also conflicting accounts of the location of the drone strike and the number of people killed. A statement by Pakistan's Ministry of Foreign Affairs on Wednesday condemned the strike and maintained that it had "targeted an Afghan refugee camp in Kurram Agency" -- an assertion that the United States rejected on Thursday. "The claim in an M.F.A. statement yesterday that U.S. forces struck an Afghan refugee camp in Kurram Agency yesterday is false," said Richard W. Snelsire, the United States Embassy spokesman in Islamabad, Pakistan's capital. American officials said that there were no Afghan refugee camps in Kurram, a remote tribal region straddling the border with Afghanistan, where they said Wednesday's drone strike had taken place.

Passively mobile communicating machines that use restricted space.

Autopilot controls are not yet fully capable of functioning without human intervention – but they're good enough to lull us into a false sense of security Wed 24 Jan 2018 03.01 EST Last modified on Wed 24 Jan 2018 03.03 EST When California police officers approached a Tesla stopped in the centre of a five-lane highway outside San Francisco last week, they found a man passed out at the wheel. The driver, who was arrested on suspicion of drunk driving, told them his car was in "autopilot", Tesla's semi-autonomous driver assist system. In a separate incident this week, firefighters in Culver City reported that a Tesla rear-ended their parked fire truck as it attended an accident on the freeway. Again, the driver stated that the vehicle was in autopilot. When u pass out behind the wheel on the Bay Bridge with more than 2x legal alcohol BAC limit and are found by a CHP Motor.

Ursula K. Le Guin, the immensely popular author who brought literary depth and a tough-minded feminist sensibility to science fiction and fantasy with books like "The Left Hand of Darkness" and the Earthsea series, died on Monday at her home in Portland, Ore. Her son, Theo Downes-Le Guin, confirmed the death. He did not specify a cause but said she had been in poor health for several months. Ms. Le Guin embraced the standard themes of her chosen genres: sorcery and dragons, spaceships and planetary conflict. But even when her protagonists are male, they avoid the macho posturing of so many science fiction and fantasy heroes.

Drone-maker DJI announced a new hobby aircraft today, one that weighs just a shade under a pound, fits in a jacket pocket, and is capable of flying itself. At that price, it hovers in DJI's lineup between the $499 DJI Spark, the gesture-controlled flyer released last year, and the more capable $999 Mavic Pro. The Mavic Air is tiny, half the size of a Mavic Pro, and about half the weight at just 15 ounces. When folded up, it's about the size of a paperback novel. At a press event in New York on Tuesday, DJI exec Michael Perry announced the Mavic Air by pulling it out of the pocket of his puffy Patagonia vest.

Google just made it a lot easier to build your very own custom AI system. A new service, called Cloud AutoML, uses several machine-learning tricks to automatically build and train a deep-learning algorithm that can recognize things in images. The technology is limited for now, but it could be the start of something big. Building and optimizing a deep neural network algorithm normally requires a detailed understanding of the underlying math and code, as well as extensive practice tweaking the parameters of algorithms to get things just right. The difficulty of developing AI systems has created a race to recruit talent, and it means that only big companies with deep pockets can usually afford to build their own bespoke AI algorithms.