Complex statistical machine learning models are increasingly being used or considered for use in high-stakes decision-making pipelines in domains such as financial services, health care, criminal justice and human services. These models are often investigated as possible improvements over more classical tools such as regression models or human judgement. While the modeling approach may be new, the practice of using some form of risk assessment to inform decisions is not. When determining whether a new model should be adopted, it is therefore essential to be able to compare the proposed model to the existing approach across a range of task-relevant accuracy and fairness metrics. Looking at overall performance metrics, however, may be misleading. Even when two models have comparable overall performance, they may nevertheless disagree in their classifications on a considerable fraction of cases. In this paper we introduce a model comparison framework for automatically identifying subgroups in which the differences between models are most pronounced. Our primary focus is on identifying subgroups where the models differ in terms of fairness-related quantities such as racial or gender disparities. We present experimental results from a recidivism prediction task and a hypothetical lending example.

Today in Entertainment: President Trump mocks'Morning Joe' hosts; Kylie and Kendall Jenner get in hot air over vintage T-shirts Kylie and Kendall Jenner wanted to sell you an old Tupac shirt for $125, but not anymore Trump's'Morning Joe' tweets rile outraged celebrities Public Enemy releases free new album on its 30th anniversary MSNBC calls out President Trump for his disparaging comments on TV hosts Seth Meyers takes on President Trump's phony Time cover'Jumanji' trailer turns Dwayne Johnson, Kevin Hart into video game avatars Trump's'Morning Joe' tweets rile outraged celebrities Seth Meyers takes on President Trump's phony Time cover'Jumanji' trailer turns Dwayne Johnson, Kevin Hart into video game avatars So perhaps it's no surprise that Public Enemy has returned with a new, free record celebrating its 30th year as a group. The group released "Nothing Is Quick in the Desert" on Thursday morning as a free download on its Bandcamp page. The record is Public Enemy's first since 2015's "Man Plans God Laughs." Public Enemy hasn't sat out the turbulent last two years in America, though. Chuck D and DJ Lord joined with most of Rage Against the Machine and Cypress Hill as the supergroup Prophets of Rage, which toured during the election season.

Uber has denied conspiring with Anthony Levandowski to steal Waymo's self-driving tech in its latest court filing. According to Bloomberg and Reuters, Uber refuted Waymo's accusation that it colluded with Levandowski to steal 14,000 files before the engineer left Google's former autonomous car division in 2015. The company vehemently denied that it hired him on the condition that he brings those files with him, or that it even knew about about the theft at all. Alphabet, Waymo's parent corporation, believes those files include the secrets of its self-driving system, including its LiDAR technology that serves as its autonomous car's eyes to see the road, obstacles and pedestrians. Those files are now the center of its lawsuit against the embattled company, which recently lost its CEO following a succession of scandals. "Prior to the filing of this lawsuit, no one at Uber knew that Levandowski had downloaded any Google proprietary information for any improper purpose or that he had deliberately taken any Google proprietary information with him when he left Google," the documents said.

Kodi has blasted sites and repositories that promote the use of illegal add-ons, and says it doesn't care if its user base drops as a result of them shutting down. The TVaddons library recently went offline with no warning, following popular Kodi add-on Phoenix, which provided access to TV shows, films and sports channels. Criminals are also selling media players pre-loaded with these add-ons, which have become known as "fully loaded Kodi boxes", despite having nothing to do with Kodi. It has now acknowledged the fact that a multitude of add-ons and repositories have started shutting down, and torn into their operators. "Due to recent legal action against websites and repositories promoting add-ons that use pirated (stolen) media content, many have shut-down their services. This is driving a large increase in users complaining in our forums and on social media about their "Kodi Box" no longer working," it wrote in a blog post.

When it comes to the EU Commission's decision to fine Google €2.42 billion for breaking competition law, the cash is just a distraction. The EU's larger goal is to position itself as an antidote to Silicon Valley's winner-takes-all attitude, and reshape our interactions with the world's tech giants. The hefty sum – the largest ever doled out by the EU's competition regulators – will sting in the short term, but Google can handle it. Alphabet, Google's parent company, made a profit of $2.5 billion (€2.2 billion) in the first six weeks of 2017 alone. The real impact of the ruling is that Google must stop using its dominance as a search engine to give itself the edge in another market: online price comparisons.

Alphabet Inc.'s most successful product -- the Google search engine -- may now be its most problematic. On Tuesday, the European Commission's top antitrust regulator levied a $2.7-billion fine against Alphabet and Google for the way the search engine handles requests for information about products. Specifically, Commissioner Margrethe Vestager said that Google skewed its results to bury links to rival companies' comparison shopping sites while prominently featuring its own service, Google Shopping. Google responded that it's simply trying to give users what they want and denied "favoring ourselves, or any particular site or seller." It has a lot at stake: Google has integrated many different offerings into its search engine, including its mapping and travel services.

There is a lot of money to be made from Artificial Intelligence. By one estimate, the market is projected to hit US$36.8 billion by 2025. Some of this money will undoubtedly go to social good, like curing illness, disease and infirmity. Some will also go to better understanding intractable social problems like wealth distribution, urban planning, smart cities, and more'efficient' ways to do just about everything. But the key word here is'some'.

The brains behind IBM's Jeopardy-winning, disease-tracking, weather-mapping Watson supercomputer plan to embark on a lobbying blitz in Washington, D.C., this week, hoping to show federal lawmakers that artificial intelligence isn't going to kill jobs -- or humans. To hear IBM tell it, much of the recent criticism around machine learning, robotics and other kinds of AI amounts to merely "fear mongering." The company's senior vice president for Watson, David Kenny, aims to convey that message to members of Congress beginning with a letter on Tuesday, stressing the "real disaster would be abandoning or inhibiting cognitive technology before its full potential can be realized." Labor experts and reams of data released in recent months argue otherwise: They foretell vast economic consequences upon the mass-market arrival of AI, as entire industries are displaced -- not just blue-collar jobs like trucking, as self-driving vehicles replace humans at the wheel, but white-collar positions like stock trading too. Others fear the privacy, security and safety implications as more tasks, from managing the country's roads to reading patients' X-ray results, are automated -- and the most dire warnings, from the likes of SpaceX and Tesla founder Elon Musk, include the potential arrival of "robots capable of destroying mankind."

Robots moving deeper into the American workplace--how much decision-making will we turn over to machines? For all the change that has come with the digital revolution – in the ways we work and communicate and do business – the real impact still lies ahead. Computers – machines themselves – are become smarter all the time. That intelligence is being wired into real world action. It's moving in on what we thought only humans could do.

Neural networks, machine-learning systems, predictive analytics, speech recognition, natural-language understanding and other components of what's broadly defined as'artificial intelligence' (AI) are currently undergoing a boom: research is progressing apace, media attention is at an all-time high, and organisations are increasingly implementing AI solutions in pursuit of automation-driven efficiencies. The first thing to establish is what we're not talking about, which is human-level AI -- often termed'strong AI' or'artificial general intelligence' (AGI). A survey conducted among four groups of experts in 2012/13 by AI researchers Vincent C. Müller and Nick Bostrom reported a 50 percent chance that AGI would be developed between 2040 and 2050, rising to 90 percent by 2075; so-called'superintelligence' -- which Bostrom defines as "any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest" -- was expected some 30 years after the achievement of AGI (Fundamental Issues of Artificial Intelligence, Chapter 33). This stuff will happen, and it certainly needs careful consideration, but it's not happening right now. What is happening right now, at an increasing pace, is the application of AI algorithms to all manner of processes that can significantly affect peoples' lives -- at work, at home and as they travel around. Although hype around these technologies is approaching the'peak of expectation' (sensu Gartner), there's a potential fly in the AI ointment: the workings of many of these algorithms are not open to scrutiny -- either because they are the proprietary assets of an organisation or because they are opaque by their very nature.