A remarkable time of human promise has been ushered in by the convergence of the ever-expanding availability of big data, the soaring speed and stretch of cloud computing platforms, and the advancement of increasingly sophisticated machine learning algorithms. Innovations in AI are already leaving a mark on government by improving the provision of essential social goods and services from healthcare, education, and transportation to food supply, energy, and environmental management. These bounties are likely just the start. The prospect that progress in AI will help government to confront some of its most urgent challenges is exciting, but legitimate worries abound. As with any new and rapidly evolving technology, a steep learning curve means that mistakes and miscalculations will be made and that both unanticipated and harmful impacts will occur. This guide, written for department and delivery leads in the UK public sector and adopted by the British Government in its publication, 'Using AI in the Public Sector,' identifies the potential harms caused by AI systems and proposes concrete, operationalisable measures to counteract them. It stresses that public sector organisations can anticipate and prevent these potential harms by stewarding a culture of responsible innovation and by putting in place governance processes that support the design and implementation of ethical, fair, and safe AI systems. It also highlights the need for algorithmically supported outcomes to be interpretable by their users and made understandable to decision subjects in clear, non-technical, and accessible ways. Finally, it builds out a vision of human-centred and context-sensitive implementation that gives a central role to communication, evidence-based reasoning, situational awareness, and moral justifiability.

Artificial intelligence doesn't need any extra fuel for the myths and misconceptions that surround it. Consider the phrase "black box" – its connotations are equal parts mysterious and ominous, the stuff of "The X Files" more than the day-to-day business of IT. Yet it's true that AI systems, such as machine learning or deep learning, take inputs and then produce outputs (or make decisions) with no decipherable explanation or context. The system makes a decision or takes some action, and we don't necessarily know why or how it arrived at that outcome. The system just does it.

Let's say that I find it curious how Spotify recommended a Justin Bieber song to me, a 40-something non-Belieber. That doesn't necessarily mean that Spotify's engineers must ensure that their algorithms are transparent and comprehensible to me; I might find the recommendation a tad off-target, but the consequences are decidedly minimal. This is a fundamental litmus test for explainable AI – that is, machine learning algorithms and other artificial intelligence systems that produce outcomes that humans can readily understand and track backwards to the origins. Conversely, relatively low-stakes AI systems might be just fine with the black box model, where we don't understand (and can't readily figure out) the results. "If algorithm results are low-impact enough, like the songs recommended by a music service, society probably doesn't need regulators plumbing the depths of how those recommendations are made," says Dave Costenaro, head of artificial intelligence R&D at Jane.ai. I can live with an app's misunderstanding of my musical tastes.

Kandinsky Figures and Kandinsky Patterns are mathematically describable, simple self-contained hence controllable test data sets for the development, validation and training of explainability in artificial intelligence. Whilst Kandinsky Patterns have these computationally manageable properties, they are at the same time easily distinguishable from human observers. Consequently, controlled patterns can be described by both humans and computers. We define a Kandinsky Pattern as a set of Kandinsky Figures, where for each figure an "infallible authority" defines that the figure belongs to the Kandinsky Pattern. With this simple principle we build training and validation data sets for automatic interpretability and context learning. In this paper we describe the basic idea and some underlying principles of Kandinsky Patterns and provide a Github repository to invite the international machine learning research community to a challenge to experiment with our Kandinsky Patterns to expand and thus make progress in the field of explainable AI and to contribute to the upcoming field of explainability and causability.

On the other hand, AM has rapidlyfrom a given document (Lippi 2016). Recent years have seen the development evolved by exploiting state-of-the-art neural architectures of a large number of techniques in this area, on coming from deep learning. So far, the wake of the advancements produced by deep these two worlds have progressed largely independently learning on the whole research field of natural of each other. Only recently, a few works language processing (NLP). Yet, it is widely recognized have taken some steps towards the integration of that the existing AM systems still have such methods, by applying techniques combining a large margin of improvement, as good results sub-symbolic classifiers with knowledge expressed have been obtained with some genres where prior in the form of rules and constraints to AM. knowledge on the structure of the text eases some Niculae et al. (2017) adopted structuredFor instance, AM tasks, but other genres such as legal cases support vector machines and recurrent neural and social media documents still require more networks to collectively classify argument components work (Cabrio and Villata, 2018). Performing and and their relations in short documents, understanding argumentation requires advanced by hard-coding contextual dependencies and constraints reasoning capabilities that are natural skills for humans, of the argument model in a factor graph. but which are difficult to learn for a machine. A joint inference approach for argument component Understanding whether a given piece of classification and relation identification was evidence supports a given claim, or whether two Persing and Ng (2016), followinginstead proposed by claims attack each other, are complex problems a pipeline scheme where integer linear programming that humans are able to address thanks to their is used to enforce mathematical constraints ability to exploit commonsense knowledge, and to on the outcomes of a first-stage set of classifiers.

Predictive models are being increasingly used to support consequential decision making at the individual level in contexts such as pretrial bail and loan approval. As a result, there is increasing social and legal pressure to provide explanations that help the affected individuals not only to understand why a prediction was output, but also how to act to obtain a desired outcome. To this end, several works have proposed methods to generate counterfactual explanations. However, they are often restricted to a particular subset of models (e.g., decision trees or linear models), and cannot directly handle the mixed (numerical and nominal) nature of the features describing each individual. In this paper, we propose a model-agnostic algorithm to generate counterfactual explanations that builds on the standard theory and tools from formal verification. Specifically, our algorithm solves a sequence of satisfiability problems, where a wide variety of predictive models and distances in mixed feature spaces, as well as natural notions of plausibility and diversity, are represented as logic formulas. Our experiments on real-world data demonstrate that our approach can flexibly handle widely deployed predictive models, while providing meaningfully closer counterfactuals than existing approaches.

Explainability issues arise because machine learning outputs are numerical; deep neural networks are so opaque that users don't necessarily know which factor contributed to what aspect of the resulting score. There are several emergent techniques for increasing explainability and interpretability of machine learning results. After organizations gain insight into the black box of intricate machine learning models, the best way to explain those results to customers, regulators and legal entities is to translate them into rules that, by their very definition, offer full transparency for explainable AI. Rules can also highlight points of bias in models.

As artificial intelligence plays an increasingly important role in our society, there are ethical and moral obligations for both businesses and researchers to ensure that their machine learning models are designed, deployed, and maintained responsibly. These models need to be rigorously audited for fairness, robustness, transparency, and interpretability. A variety of methods have been developed that focus on these issues in isolation, however, managing these methods in conjunction with model development can be cumbersome and timeconsuming. In this paper, we introduce a unified and model-agnostic approach to address these issues: Counterfactual Explanations for Robustness, Transparency, Interpretability, and Fairness of Artificial Intelligence models (CERTIFAI). Unlike previous methods in this domain, CERTIFAI is a general tool that can be applied to any black-box model and any type of input data. Given a model and an input instance, CERTIFAI uses a custom genetic algorithm to generate counterfactuals: instances close to the input that change the prediction of the model. We demonstrate how these counterfactuals can be used to examine issues of robustness, interpretability, transparency, and fairness. Additionally, we introduce CERScore, the first black-box model robustness score that performs comparably to methods that have access to model internals.

Post-hoc explanations of machine learning models are crucial for people to understand and act on algorithmic predictions. An intriguing class of explanations is through counterfactuals, hypothetical examples that show people how to obtain a different prediction. We posit that effective counterfactual explanations should satisfy two properties: feasibility of the counterfactual actions given user context and constraints, and diversity among the counterfactuals presented. To this end, we propose a framework for generating and evaluating a diverse set of counterfactual explanations based on average distance and determinantal point processes. To evaluate the actionability of counterfactuals, we provide metrics that enable comparison of counterfactual-based methods to other local explanation methods. We further address necessary tradeoffs and point to causal implications in optimizing for counterfactuals. Our experiments on three real-world datasets show that our framework can generate a set of counterfactuals that are diverse and well approximate local decision boundaries.

Explainable AI is the idea that an AI algorithm should be able to explain how it reached a conclusion in a way that humans can decipher. There's a well-known "black box problem" where an AI algorithm can determine something but is not able to give details about the factors that caused that result. Then, how can people feel that they can authoritatively trust what an AI algorithm says? Getting to the goal of explainable AI is an admirable and necessary feat, but it's not straightforward. Here are four things that the tech industry needs to realize before we get to the point of making explainable AI a reality. It's understandable why people assert that if AI can explain itself, members of the general public, as well as businesses who use AI to make decisions, will believe it's more trustworthy.