Paul Duan was working as a data scientist at Eventbrite in San Francisco by day, and volunteering at homeless shelters and soup kitchens by night. He realized one day that he wanted to use AI to help unemployed people find jobs--a core mission of his Paris, France-based nonprofit Bayes Impact. Bayes Impact uses data to build social services fit for a better future. "When you work at a soup kitchen, you serve a soup one by one to each individual, and it feels great," says Duan, "But then it gets really sad because you see that there are 50 people in line behind the person, and you know that behind the closed door of the shelter you have 10,000 more on the streets. So the one question that came to mind was, 'how can we impact people at the biggest scale?'"

A central goal of algorithmic fairness is to reduce bias in automated decision making. An unavoidable tension exists between accuracy gains obtained by using sensitive information (e.g., gender or ethnic group) as part of a statistical model, and any commitment to protect these characteristics. Often, due to biases present in the data, using the sensitive information in the functional form of a classifier improves classification accuracy. In this paper we show how it is possible to get the best of both worlds: optimize model accuracy and fairness without explicitly using the sensitive feature in the functional form of the model, thereby treating different individuals equally. Our method is based on two key ideas. On the one hand, we propose to use Multitask Learning (MTL), enhanced with fairness constraints, to jointly learn group specific classifiers that leverage information between sensitive groups. On the other hand, since learning group specific models might not be permitted, we propose to first predict the sensitive features by any learning method and then to use the predicted sensitive feature to train MTL with fairness constraints. This enables us to tackle fairness with a three-pronged approach, that is, by increasing accuracy on each group, enforcing measures of fairness during training, and protecting sensitive information during testing. Experimental results on two real datasets support our proposal, showing substantial improvements in both accuracy and fairness.

Recent studies on disparate impact in machine learning applications have sparked a debate around the concept of fairness along with attempts to formalize its different criteria. Many of these approaches focus on reducing prediction errors while maximizing sole utility of the institution. This work seeks to reconceptualize and critically frame the existing discourse on fairness by underlining the implicit biases embedded in common understandings of fairness in the literature and how they contrast with its corresponding economic and legal definitions. This paper expands the concept of utility and fairness by bringing in concepts from established literature in welfare economics and game theory. We then translate these concepts for the algorithmic prediction domain by defining a formalization of Nash Welfare Product that seeks to expand utility by collapsing that of the institution using the prediction tool and the individual subject to the prediction into one function. We then apply a modulating function that makes the fairness and welfare trade-offs explicit based on designated policy goals and then apply it to a temporal model to take into account the effects of decisions beyond the scope of one-shot predictions. We apply this on a binary classification problem and present results of a multi-epoch simulation based on the UCI Adult Income dataset and a test case analysis of the ProPublica recidivism dataset that show that expanding the concept of utility results in a fairer distribution correcting for the embedded biases in the dataset without sacrificing the classifier accuracy.

This is a rush transcript from "The Five," October 17, 2018. This copy may not be in its final form and may be updated. It's 5 o'clock in New York City, and this is The Five. The liberal media is once again whipping itself into a frenzy over President Trump. First up, amid the presidents brewing battle with Stormy Daniels and her lawyer, Michael Avenatti, Trump-hating MSNBC host Mika Brzezinski is calling on the president to be removed from office. MIKA BRZEZINSKI, MSNBC: This is one of the many, many, many ways this president has shown us that he's not fit, possibly not even well. You're working for a president who is not fit to lead, who's going to do something crazy in 5 minutes, one hour, tonight or tomorrow. Like what more do you need to hear from him to start thinking 25th amendment or something else? DON LEMON, CNN: Does he own a mirror? Has he -- he keeps talking about people gaining weight and how people look? Has he -- does he own a mirror that doesn't have Vaseline over it or a cloth? I mean, all he has to do is look in the mirror. Donald Trump is no prize. And if I were him, not that I'm one either, I would keep my thoughts about other people's looks to myself. Some in the media are trying to spin Elizabeth Warren's disastrous DNA reveal by using it to attack Trump. It is ultimately a dog whistle that plays into the grievances of his base, his overwhelmingly white bass, and it goes into multiple themes that are at issue for conservatives, predominately around affirmative action and whether or not they're people who are sort of cheating the system by claiming to be minorities. WATTERS: And the architect of the Iran nuclear deal, former Obama adviser, Ben Rhodes, is parroting this new media talking point about the disappearance of the Washington Post columnist. BEN RHODES, FORMER OBAMA OFFICIAL: The message -- the Saudis wanted to send and they have sent is that you're not safe anywhere if you criticize us. And the message of President Trump is sending is that there's no consequences. We have a President of the United States who says Journalist (INAUDIBLE). So values like freedom of speech and dissent, suddenly are very endangered around the world. And that's a thread line that I think it's getting much worse. Juan, let's pick up on what Ben Rhodes just said. I think it's pretty irresponsible to link the Washington Post columnist death with President Trump's war on the media.

The relationships are predicted from local polynomial regressions. Shaded areas indicate 95% confidence intervals. Preferences concerning time, risk, and social interactions systematically shape human behavior and contribute to differential economic and social outcomes between women and men. We present a global investigation of gender differences in six fundamental preferences. Our data consist of measures of willingness to take risks, patience, altruism, positive and negative reciprocity, and trust for 80,000 individuals in 76 representative country samples. Gender differences in preferences were positively related to economic development and gender equality. This finding suggests that greater availability of and gender-equal access to material and social resources favor the manifestation of gender-differentiated preferences across countries. Fundamental preferences such as altruism, risk-taking, reciprocity, patience, or trust constitute the foundation of choice theories and govern human behavior.

Big Data is all the rage today, but Small Data matters too! Drawing reliable conclusions from small datasets, like those from clinical trials for rare diseases or in studies of endangered species, remains one of the trickiest obstacles in statistics. Now, Cold Spring Harbor Laboratory (CSHL) researchers have developed a new way to analyze small data, one inspired by advanced methods in theoretical physics, but available as easy-to-use software. "Dealing with small datasets is a fundamental part of doing science," CSHL Assistant Professor Justin Kinney explained. The challenge is that, with very little data, it's not only hard to come to a conclusion; it's also hard to determine how certain your conclusions are.

Today, Above the Law and Thomson Reuters present Big Data and the Litigation Analytics Revolution, the fourth and final installment of our Law2020 series, a multimedia exploration of how artificial intelligence and other cutting-edge technologies are reshaping the practice and profession of law. Previous Law2020 articles have explored the implications of AI for legal education, legal ethics, and legal research. Today, we take a deep dive into how sophisticated litigators are leveraging Big Data and analytics to decide critical questions of case strategy and tactics. Additionally, we explore litigation analytics is empowering lawyers to excel on many other fronts, including managing client expectations, accelerating client service, refining law firm operations, and optimizing legal research. You can read the feature here, and you can sign up using the form below to learn more information about Law2020.

An American-led tech consortium dedicated to safeguarding the development of artificial intelligence has welcomed its first Chinese member, internet search company Baidu. The Partnership on AI (PAI) was set up two years ago to generate best practices for AI technology. It's funded by its members, which include companies like Google, Amazon, Facebook, and Microsoft, and also partners with government entities like the UN and Human Rights Watch. Membership does not necessitate any legally binding promises, but companies who join PAI must "believe in and endeavor to uphold" eight key tenets. PAI's executive director, Terah Lyons, told The Verge that the group cannot accomplish its aims without "insight from the leading global AI actors" -- including Chinese firms.

Big Data is all the rage today, but Small Data matters too! Drawing reliable conclusions from small datasets, like those from clinical trials for rare diseases or in studies of endangered species, remains one of the trickiest obstacles in statistics. Now, Cold Spring Harbor Laboratory (CSHL) researchers have developed a new way to analyze small data, one inspired by advanced methods in theoretical physics, but available as easy-to-use software. "Dealing with small datasets is a fundamental part of doing science," CSHL Assistant Professor Justin Kinney explained. The challenge is that, with very little data, it's not only hard to come to a conclusion; it's also hard to determine how certain your conclusions are.

MORRISTOWN, N.J.--(BUSINESS WIRE)--The digital era shift is realigning fundamental elements of business that require major adjustments from insurers in order for them to survive and thrive, according to a new thought leadership report released today by Majesco (NYSE AMERICAN:MJCO), a global provider of core insurance platform software and consulting services for insurance business transformation. A new digital era of insurance focused on innovation and growth requires platform-based business models and solutions to succeed. The report, A New Business Model for a New Era of Insurance: Digital Insurance 2.0 in the Platform and API Economy, highlights the impacts of the shift to a new age of insurance, Digital Insurance 2.0. Underpinning this new era is the shift to the application programming interface ("API") and platform economy, which consists of key technologies such as cloud computing, open APIs, microservices, ecosystems, and data and analytics, which together help insurers create new business models, products and services, connect everything and, most importantly, create new customer experiences. "The digital age shift is now top of mind for every leadership team and board because it is extending an organization's growth and innovation capabilities," remarked Denise Garth, SVP - Strategic Marketing, Industry Relations and Innovation for Majesco.