This year in science saw important developments, including a few surprises--but, many of 2018's most significant events were the products of ongoing social and political trends that have been in motion for years. The year 2018 saw renewed debate about the urgency and efficacy of policies to limit climate change, against a backdrop of severe weather events and at a time when the U.S. government continues to roll back environmental regulation. It was also a year when both CRISPR and artificial intelligence (AI) revealed their growing potential for innovation across many fields, even as researchers expressed concern that these technologies were developing at speeds too fast to assess and evaluate their impact. NASA continued its exploration of the solar system, and Americans had the opportunity to watch, in real time, the harrowing landing of the InSight probe on Mars. But, the public was also fixated on the launches of SpaceX's revolutionary, reusable rockets--the latest chapter in the rise of private space companies, as the Trump administration continues to look for ways to outsource and commercialize U.S. space exploration.

The video showed the woman in a pink off-the-shoulder top, sitting on a bed, smiling a convincing smile. But it had been seamlessly grafted, without her knowledge or consent, onto someone else's body: a young pornography actress, just beginning to disrobe for the start of a graphic sex scene. A crowd of unknown users had been passing it around online. She felt nauseous and mortified: What if her co-workers saw it? Would it change how they thought of her? Would they believe it was a fake?

Google Photos users nervous about facial recognition on the service aren't going to be very happy. A Chicago judge has granted Google a motion dismissing a lawsuit accusing the company of violating Illinois' Biometric Information Privacy Act by gathering biometric data from photos without permission. The plaintiffs couldn't demonstrate that they'd suffered "concrete injuries" from the facial recognition system, according to the judge. The suit had been filed in March 2016. The complainants wanted over $5 million for state residents, with $5,000 for each purposeful violation and $1,000 for each unintentional violation.

Einat Weiss, VP Global Marketing at NICE, writes about how marketers can use Artificial Intelligence (AI) for brand building. As we approach 2019, it's the perfect time to reflect on where the marketing industry has been in the past year and where it's headed. In 2018, we saw more organizations relying heavily on AI-driven messaging applications and social media to engage with customers. In fact, 72 percent of millennials believe a phone call is not the best way to resolve their customer services issues. Additionally, experts predict that by 2020, 85 percent of all customer service interactions will be handled without the need for a human agent.

Interest in artificial intelligence technology, including in the association space, is picking up in a big way as we close out 2018 and move into the new year, and an array of recent studies on the topic shows just what we have to look forward to. Studies from Stanford University, Pew Research Center, and New York University examine the state of AI from different directions. They explore the field's global growth, its diversity, and its economic and ethical implications. As highlighted by the fact that numerous reports on the topic came out at around the same time, artificial intelligence is driving a lot of research momentum these days. According to Stanford's AI Index [PDF], growth in the number of artificial intelligence papers published in a given year has outpaced that of even computer science.

Societies often rely on human experts to take a wide variety of decisions affecting their members, from jail-or-release decisions taken by judges and stop-and-frisk decisions taken by police officers to accept-or-reject decisions taken by academics. In this context, each decision is taken by an expert who is typically chosen uniformly at random from a pool of experts. However, these decisions may be imperfect due to limited experience, implicit biases, or faulty probabilistic reasoning. Can we improve the accuracy and fairness of the overall decision making process by optimizing the assignment between experts and decisions? In this paper, we address the above problem from the perspective of sequential decision making and show that, for different fairness notions from the literature, it reduces to a sequence of (constrained) weighted bipartite matchings, which can be solved efficiently using algorithms with approximation guarantees. Moreover, these algorithms also benefit from posterior sampling to actively trade off exploitation---selecting expert assignments which lead to accurate and fair decisions---and exploration---selecting expert assignments to learn about the experts' preferences and biases. We demonstrate the effectiveness of our algorithms on both synthetic and real-world data and show that they can significantly improve both the accuracy and fairness of the decisions taken by pools of experts.

The Equalized Odds (for short, EO) is one of the most popular measures of discrimination used in the supervised learning setting. It ascertains fairness through the balance of the misclassification rates (false positive and negative) across the protected groups -- e.g., in the context of law enforcement, an African-American defendant who would not commit a future crime will have an equal opportunity of being released, compared to a non-recidivating Caucasian defendant. Despite this noble goal, it has been acknowledged in the literature that statistical tests based on the EO are oblivious to the underlying causal mechanisms that generated the disparity in the first place (Hardt et al. 2016). This leads to a critical disconnect between statistical measures readable from the data and the meaning of discrimination in the legal system, where compelling evidence that the observed disparity is tied to a specific causal process deemed unfair by society is required to characterize discrimination. The goal of this paper is to develop a principled approach to connect the statistical disparities characterized by the EO and the underlying, elusive, and frequently unobserved, causal mechanisms that generated such inequality. We start by introducing a new family of counterfactual measures that allows one to explain the misclassification disparities in terms of the underlying mechanisms in an arbitrary, non-parametric structural causal model. This will, in turn, allow legal and data analysts to interpret currently deployed classifiers through causal lens, linking the statistical disparities found in the data to the corresponding causal processes. Leveraging the new family of counterfactual measures, we develop a learning procedure to construct a classifier that is statistically efficient, interpretable, and compatible with the basic human intuition of fairness. We demonstrate our results through experiments in both real (COMPAS) and synthetic datasets.

We study the interplay between sequential decision making and avoiding discrimination against protected groups, when examples arrive online and do not follow distributional assumptions. We consider the most basic extension of classical online learning: Given a class of predictors that are individually non-discriminatory with respect to a particular metric, how can we combine them to perform as well as the best predictor, while preserving non-discrimination? Surprisingly we show that this task is unachievable for the prevalent notion of "equalized odds" that requires equal false negative rates and equal false positive rates across groups. On the positive side, for another notion of non-discrimination, "equalized error rates", we show that running separate instances of the classical multiplicative weights algorithm for each group achieves this guarantee. Interestingly, even for this notion, we show that algorithms with stronger performance guarantees than multiplicative weights cannot preserve non-discrimination.

Understanding how humans perceive the likability of high-dimensional ``objects'' such as faces is an important problem in both cognitive science and AI/ML. Existing models generally assume these preferences to be fixed. However, psychologists have found human assessment of facial attractiveness to be context-dependent. Specifically, the classical Beauty-in-Averageness (BiA) effect, whereby a blended face is judged to be more attractive than the originals, is significantly diminished or reversed when the original faces are recognizable, or when the blend is mixed-race/mixed-gender and the attractiveness judgment is preceded by a race/gender categorization, respectively. This "Ugliness-in-Averageness" (UiA) effect has previously been explained via a qualitative disfluency account, which posits that the negative affect associated with the difficult race or gender categorization is inadvertently interpreted by the brain as a dislike for the face itself. In contrast, we hypothesize that human preference for an object is increased when it incurs lower encoding cost, in particular when its perceived {\it statistical typicality} is high, in consonance with Barlow's seminal ``efficient coding hypothesis.'' This statistical coding cost account explains both BiA, where facial blends generally have higher likelihood than ``parent faces'', and UiA, when the preceding context or task restricts face representation to a task-relevant subset of features, thus redefining statistical typicality and encoding cost within that subspace. We use simulations to show that our model provides a parsimonious, statistically grounded, and quantitative account of both BiA and UiA. We validate our model using experimental data from a gender categorization task. We also propose a novel experiment, based on model predictions, that will be able to arbitrate between the disfluency account and our statistical coding cost account of attractiveness.

Inspired by "predictive coding" - a theory in neuroscience, we develop a bi-directional and dynamic neural network with local recurrent processing, namely predictive coding network (PCN). Unlike feedforward-only convolutional neural networks, PCN includes both feedback connections, which carry top-down predictions, and feedforward connections, which carry bottom-up errors of prediction. Feedback and feedforward connections enable adjacent layers to interact locally and recurrently to refine representations towards minimization of layer-wise prediction errors. When unfolded over time, the recurrent processing gives rise to an increasingly deeper hierarchy of non-linear transformation, allowing a shallow network to dynamically extend itself into an arbitrarily deep network. We train and test PCN for image classification with SVHN, CIFAR and ImageNet datasets. Despite notably fewer layers and parameters, PCN achieves competitive performance compared to classical and state-of-the-art models. Further analysis shows that the internal representations in PCN converge over time and yield increasingly better accuracy in object recognition. Errors of top-down prediction also reveal visual saliency or bottom-up attention.