Interpretability has become an important topic of research as more machine learning (ML) models are deployed and widely used to make important decisions. Due to it's complexity, i For high-stakes domains such as medical, providing intuitive explanations that can be consumed by domain experts without ML expertise becomes crucial. To this demand, concept-based methods (e.g., TCAV) were introduced to provide explanations using user-chosen high-level concepts rather than individual input features. While these methods successfully leverage rich representations learned by the networks to reveal how human-defined concepts are related to the prediction, they require users to select concepts of their choice and collect labeled examples of those concepts. In this work, we introduce DTCAV (Discovery TCAV) a global concept-based interpretability method that can automatically discover concepts as image segments, along with each concept's estimated importance for a deep neural network's predictions. We validate that discovered concepts are as coherent to humans as hand-labeled concepts. We also show that the discovered concepts carry significant signal for prediction by analyzing a network's performance with stitched/added/deleted concepts. DTCAV results revealed a number of undesirable correlations (e.g., a basketball player's jersey was a more important concept for predicting the basketball class than the ball itself) and show the potential shallow reasoning of these networks.

People are rated and ranked, towards algorithmic decision making in an increasing number of applications, typically based on machine learning. Research on how to incorporate fairness into such tasks has prevalently pursued the paradigm of group fairness: giving adequate success rates to specifically protected groups. In contrast, the alternative paradigm of individual fairness has received relatively little attention, and this paper advances this less explored direction. The paper introduces a method for probabilistically mapping user records into a low-rank representation that reconciles individual fairness and the utility of classifiers and rankings in downstream applications. Our notion of individual fairness requires that users who are similar in all task-relevant attributes such as job qualification, and disregarding all potentially discriminating attributes such as gender, should have similar outcomes. We demonstrate the versatility of our method by applying it to classification and learning-to-rank tasks on a variety of real-world datasets. Our experiments show substantial improvements over the best prior work for this setting.

As we enter 2019, it's more clear than ever that data isn't going to simply play a supporting role – not for any industry, and certainly not for recruitment, where it may well take center stage. Artificial intelligence, machine learning, scientific approaches to assessment, talent intelligence, and other areas are dependent on it, so to keep up, recruitment firms must have a wealth of actionable information to draw on. To do so effectively, they must appoint data-driven business leaders. Just as many firms have a C-Suite, modern recruitment firms should have a D-Suite – where analyzing information and using insight to improve key business processes are treated as urgently as sourcing candidates; pleasing clients while fulfilling other critical business functions. If you're running a recruitment firm, here are three reasons to appoint some data-driven leaders.

On March 23rd 2016, Microsoft released a new artificial intelligence Twitter bot named Tay. "Hellooooooo world!!!" read its cutesy first message. Within hours, however, human users had persuaded Tay to replace its light hearted banter with anti-semitic, sexist, and racist Tweets. The media got hold of the story and pilloried Microsoft and its new CEO, Satya Nadella. While it probably didn't feel like it at the time, Tay represented the start of a significant turnaround in Microsoft's fortunes that would eventually lead the tech giant to reclaim its position as the most valuable company in the world.

Artificial intelligence has started -- slowly -- to make its presence felt in payments and commerce, including in fraud prevention, via early deployments of the technology and cutting-edge AI algorithms. And with those deployments comes increasing awareness of what AI can really do, how it can improve upon less sophisticated machine learning technology, and why it promises to play a vital role in the daily lives of consumers in the coming decades. The race to get ahead on the technology is now gaining clarity as well. Fresh data from the U.N. World Intellectual Property Organization, or WIPO, finds that the U.S. and China are building global dominance in AI technology development (along with closely related tech that is finding more use among financial institutions). The study is based on "more than 340,000 AI-related patent applications and 1.6 million scientific papers published since AI first emerged in the 1950s, with the majority of all AI-related patent filings published since 2013."

Machine learning models or algorithms have shown over the past few years that they can exhibit human traits like racism and sexism by misidentifying black people as gorillas (Barr, 2015) or perpetuating gender income inequality through ad suggestions (Datta et al., 2015). Algorithmic bias, however, is not inherently problematic. Given the potential harm machine learning can cause, how can South African organisations mitigate against problematic algorithmic bias in their data and models? This essay will use the taxonomy of algorithmic bias created by Danks and London (2017) to differentiate between the various types of algorithmic bias and give examples of how problematic bias might perpetuate immoral discrimination within a South African context. Specifically, it will examine training data bias, algorithmic focus bias and transfer context bias. The most intuitive bias is training data bias; if biased data are used, the resulting model reflects that bias.

The Australian Human Rights Commission is conducting a project on Human Rights and New Technology (the Project). As part of the Project, the Commission and the World Economic Forum are working together to explore models of governance and leadership on artificial intelligence (AI) in Australia. This White Paper has been produced to support a consultation process that aims to identify how Australia can simultaneously foster innovation and protect human rights – as we see unprecedented growth in new technologies, such as AI. The White Paper complements the broader issues raised in the Commission's Human Rights and Technology Issues Paper. The consultation conducted on the Issues Paper and White Paper will inform the Commission's proposals for reform, to be released in mid-2019. The White Paper asks whether Australia needs an organisation to take a central role in promoting responsible innovation in AI and related technology and, if so, what that organisation could look like.

Facial recognition technology is everywhere you look -- from unlocking phones to shaming jaywalkers. But should corporations have the power to use it on you without consent? That's the question the city of San Francisco is tackling right now. A member of the city's Board of Supervisors proposed a ban on facial recognition technology for city agencies on Tuesday, Wired reports -- potentially forcing tech companies to justify the use of surveillance tools. San Francisco city board member Aaron Peskin is calling for an approval process for any new surveillance technology purchases by city agencies such as license plate readers, CCTV, and gun-detection systems. "I have yet to be persuaded that there is any beneficial use of this technology that outweighs the potential for government actors to use it for coercive and oppressive ends," Peskin told Wired.

This is an integrative review that address the question, "What makes for a good explanation?" with reference to AI systems. Pertinent literatures are vast. Thus, this review is necessarily selective. That said, most of the key concepts and issues are expressed in this Report. The Report encapsulates the history of computer science efforts to create systems that explain and instruct (intelligent tutoring systems and expert systems). The Report expresses the explainability issues and challenges in modern AI, and presents capsule views of the leading psychological theories of explanation. Certain articles stand out by virtue of their particular relevance to XAI, and their methods, results, and key points are highlighted. It is recommended that AI/XAI researchers be encouraged to include in their research reports fuller details on their empirical or experimental methods, in the fashion of experimental psychology research reports: details on Participants, Instructions, Procedures, Tasks, Dependent Variables (operational definitions of the measures and metrics), Independent Variables (conditions), and Control Conditions.

AI intensive systems that operate upon user data face the challenge of balancing data utility with privacy concerns. We propose the idea and present the prototype of an open-source tool called Privacy Utility Trade-off (PUT) Workbench which seeks to aid software practitioners to take such crucial decisions. We pick a simple privacy model that doesn't require any background knowledge in Data Science and show how even that can achieve significant results over standard and real-life datasets. The tool and the source code is made freely available for extensions and usage.