Google DeepMind and the NHS are developing a machine learning system with Moorfields Eye Hospital that can recognize sight-threatening conditions from just a digital scan of the eye. Mustafa Suleyman, Deepmind's co-founder, says this is the company's first foray into a purely medical research. In this new collaboration with Moorfields, an algorithm will be trained using one million anonymized eye scans to train to identify early signs of degenerative eye conditions such as wet age-related macular degeneration and diabetic retinopathy. "If you have diabetes you're 25 times more likely to go blind. If we can detect this, and get in there as early as possible, then 98% of the most severe visual loss might be prevented," says Suleyman.

The Obama administration wants the U.S. to be a world leader in economic and defense policy related to a new wave of automation powered by machine learning and artificial intelligence, White House Chief of Staff Denis McDonough said Tuesday. "We can return to these questions in a way that America can kind of set the space and then set the parameters for how we go about it," McDonough said during a White House conversation on the topic that he moderated. The conversation comes as the White House looks to wrap up its string of workshops on artificial intelligence Thursday. In May, the White House Office of Science and Technology Policy announced plans to explore the uses and risks of AI. Since then the office has hosted three workshops and another event on the matter.

We investigate a long-debated question, which is how to create predictive models of recidivism that are sufficiently accurate, transparent, and interpretable to use for decision-making. This question is complicated as these models are used to support different decisions, from sentencing, to determining release on probation, to allocating preventative social services. Each use case might have an objective other than classification accuracy, such as a desired true positive rate (TPR) or false positive rate (FPR). Each (TPR, FPR) pair is a point on the receiver operator characteristic (ROC) curve. We use popular machine learning methods to create models along the full ROC curve on a wide range of recidivism prediction problems. We show that many methods (SVM, Ridge Regression) produce equally accurate models along the full ROC curve. However, methods that designed for interpretability (CART, C5.0) cannot be tuned to produce models that are accurate and/or interpretable. To handle this shortcoming, we use a new method known as SLIM (Supersparse Linear Integer Models) to produce accurate, transparent, and interpretable models along the full ROC curve. These models can be used for decision-making for many different use cases, since they are just as accurate as the most powerful black-box machine learning models, but completely transparent, and highly interpretable.

Few environmental limits are as obvious to people today as water availability. Particularly in drier climates, availability can be a pretty unforgiving equation. Even there, a family might pay less for water than for cell phones, but there is often a pretty complex system behind your tap that keeps it running. The challenge of water availability rises beyond engineering. It becomes a delicate dance managing demand, forecasting supply, and sustaining ecosystems. Decisions have to be made based on information that is never complete, so any opportunity to obtain more useful information is liable to get a thirsty look from water managers.

The United States Air Force wants robots and service members to be best buds on the battlefield. Last Friday, the Air Force announced a grant of 7.5 million for research on ways to make humans trust artificial intelligence (AI) so that people and machines can collaborate on missions. Soon service members in every branch of the Armed Forces will be working with AI on a daily basis--be it unmanned ariel vehicles, underwater drones, or robot soldiers (the U.S. Military had to shelve the Boston Dynamic L3 "robotic mule" because it was too loud, but last week the company revealed the much stealthier SpotMini). On November 1, 2014, (one week after Elon Musk compared developing AI to "summoning the demon") Undersecretary of Defense Frank Kendall issued a memo asking the Defense Science Board to study what issues must be solved in order to expand the use of AI "across all war-fighting domains." But robotic weapons and soldiers wont be as effective if their human counterparts don't trust them.

On June 27, the US Office of Science and Technology Policy (OSTP) issued a Request For Information on Artificial Intelligence (RFI). The RFI seeks public input on the tools, technologies, and training needed to further research and implementation of artificial intelligence (AI). Public feedback on these important questions will enable the OSTP to develop guidance on how the law should balance AI's potential benefits with its numerous threats. For example, self-driving cars could improve driving safety and provide mobility for people with disabilities. Likewise, the use of AI in healthcare has the potential to dramatically improve the quality and accuracy of medical care.

A judge in Connecticut Wednesday said he planned to rule within a week in a father and son's case against the Federal Aviation Administration over YouTube videos of gun-toting, flame-throwing drones. Austin Haughwout, 19, of Clinton, and his father, Bret Haughwout, produced the videos. They've refused to comply with subpoenas issued by the U.S. attorney's office on behalf of the FAA, saying the subpoenas violate their constitutional right to be free from unreasonable searches and seizures and questioning the agency's authority to regulate recreational drones. U.S. District Judge Jeffrey Meyer gave both sides a deadline of Monday, July 11 to file any additional documents. One of the Haughwouts' videos, viewed more than 3.7 million times, shows a flying drone equipped with a handgun firing rounds.

SUMMARY: Artificial intelligence (AI) technologies offer great promise for creating new and innovative products, growing the economy, and advancing national priorities in areas such as education, mental and physical health, addressing climate change, and more. Like any transformative technology, however, AI carries risks and presents complex policy challenges along a number of different fronts. The Office of Science and Technology Policy (OSTP) is interested in developing a view of AI across all sectors for the purpose of recommending directions for research and determining challenges and opportunities in this field. The views of the American people, including stakeholders such as consumers, academic and industry researchers, private companies, and charitable foundations, are important to inform an understanding of current and future needs for AI in diverse fields. The purpose of this RFI is to solicit feedback on overarching questions in AI, including AI research and the tools, technologies, and training that are needed to answer these questions.

A park ranger treads carefully through the trees, stopping to listen for signs of the poacher he's tailing. Killed for skins, medicine and trophy hunting, the worldwide population of tigers has been reduced to near-extinction at about 3,200. The scale of destruction is increasing, and it will take a three-pronged approach to battle the corruption and financial incentives driving the illegal trade: tackling the source, transmission and demand for wild animal products. Supply could be dealt with by park rangers catching the poachers before they attack, but finding a single poacher in thousands of square kilometers can be almost impossible, and in the poorest areas, resources are so constrained that poachers are not being intercepted at all. Artificial intelligence and game theory are the surprising elements in the arsenal of weapons used to combat this problem.

Across the world there is an estimated 285 million visually impaired people, and 39 million of these are blind. Conditions like age-related macular degeneration and diabetic retinopathy can be picked up is using digital screenings, which are highly complex and take a lot of time to analyse. Now Google's DeepMind Health is teaming up with a London eye hospital to investigate how machine learning could help analyse these scans efficiently and effectively. Moorfields Eue Hospital in London has announced a new medical research partnership with Google's DeepMind Health that could revolutionise the way professionals carry out eye tests and lead to earlier detection of common eye diseases Diabetes is on the rise. It's estimated that 1 in 11 of the world's adult population are affected.