Recent works have shown that selecting an optimal model architecture suited to the differential privacy setting is necessary to achieve the best possible utility for a given privacy budget using differentially private stochastic gradient descent (DP-SGD)(Tramer and Boneh 2020; Cheng et al. 2022). In light of these findings, we empirically analyse how different fairness notions, belonging to distinct classes of statistical fairness criteria (independence, separation and sufficiency), are impacted when one selects a model architecture suitable for DP-SGD, optimized for utility. Using standard datasets from ML fairness literature, we show using a rigorous experimental protocol, that by selecting the optimal model architecture for DP-SGD, the differences across groups concerning the relevant fairness metrics (demographic parity, equalized odds and predictive parity) more often decrease or are negligibly impacted, compared to the non-private baseline, for which optimal model architecture has also been selected to maximize utility. These findings challenge the understanding that differential privacy will necessarily exacerbate unfairness in deep learning models trained on biased datasets.

Many stakeholders struggle to make reliances on ML-driven systems due to the risk of harm these systems may cause. Concerns of trustworthiness, unintended social harms, and unacceptable social and ethical violations undermine the promise of ML advancements. Moreover, such risks in complex ML-driven systems present a special challenge as they are often difficult to foresee, arising over periods of time, across populations, and at scale. These risks often arise not from poor ML development decisions or low performance directly but rather emerge through the interactions amongst ML development choices, the context of model use, environmental factors, and the effects of a model on its target. Systems safety engineering is an established discipline with a proven track record of identifying and managing risks even in high-complexity sociotechnical systems. In this work, we apply a state-of-the-art systems safety approach to concrete applications of ML with notable social and ethical risks to demonstrate a systematic means for meeting the assurance requirements needed to argue for safe and trustworthy ML in sociotechnical systems.

This paper considers doing quantile regression on censored data using neural networks (NNs). This adds to the survival analysis toolkit by allowing direct prediction of the target variable, along with a distribution-free characterisation of uncertainty, using a flexible function approximator. We begin by showing how an algorithm popular in linear models can be applied to NNs. However, the resulting procedure is inefficient, requiring sequential optimisation of an individual NN at each desired quantile. Our major contribution is a novel algorithm that simultaneously optimises a grid of quantiles output by a single NN. To offer theoretical insight into our algorithm, we show firstly that it can be interpreted as a form of expectation-maximisation, and secondly that it exhibits a desirable `self-correcting' property. Experimentally, the algorithm produces quantiles that are better calibrated than existing methods on 10 out of 12 real datasets.

Without a human-centric approach, OpenAI ChatGPT runs on the data available on the various channels, which can also deliver services without meeting the context requirements. Sometimes, it writes plausible-sounding content but can be trustworthy. The new kid on the block, AI-powered ChatGPT offers numerous exceptional services and is claimed to be useful for coding, content writing, etc., minimizing human intervention. As erudite machinery becomes a trending sensation, companies can also see AI biases, security risks, and less personalized CX. The uncapped accessibility, and unrestricted usage of ChatGPT have increased the cybersecurity risks that can hamper the whole organization. Through ChatGPT, cybercriminals can draft a fraudulent email carrying unsecured links, attachments providing sensitive data, or instructions regarding transferring money into specific accounts from a reputed company or person.

Offering no criteria by which to define unacceptable risk for AI systems and no method to add new high-risk applications to the Act if such applications are discovered to pose a substantial danger of harm. This is particularly problematic because AI systems are becoming broader in their utility. Only requiring that companies take into account harm to individuals, excluding considerations of indirect and aggregate harms to society. An AI system that has a very small effect on, e.g., each person's voting patterns might in the aggregate have a huge social impact. Permitting virtually no public oversight over the assessment of whether AI meets the Act's requirements.

A new Rembrandt painting – by a computer. Photos of people who don’t exist: when it comes to realizing art world fantasies, Artificial Intelligence is delivering the goods. A great opportunity - or are there dangers? How AI is changing creativity.

In this paper we propose a new approach to detect clusters in undirected graphs with attributed vertices. We incorporate structural and attribute similarities between the vertices in an augmented graph by creating additional vertices and edges as proposed in [1, 2]. The augmented graph is then embedded in a Euclidean space associated to its Laplacian and we cluster vertices via a modified K-means algorithm, using a new vector-valued distance in the embedding space. Main novelty of our method, which can be classified as an early fusion method, i.e., a method in which additional information on vertices are fused to the structure information before applying clustering, is the interpretation of attributes as new realizations of graph vertices, which can be dealt with as coordinate vectors in a related Euclidean space. This allows us to extend a scalable generalized spectral clustering procedure which substitutes graph Laplacian eigenvectors with some vectors, named algebraically smooth vectors, obtained by a linear-time complexity Algebraic MultiGrid (AMG) method. We discuss the performance of our proposed clustering method by comparison with recent literature approaches and public available results. Extensive experiments on different types of synthetic datasets and real-world attributed graphs show that our new algorithm, embedding attributes information in the clustering, outperforms structure-only-based methods, when the attributed network has an ambiguous structure. Furthermore, our new method largely outperforms the method which originally proposed the graph augmentation, showing that our embedding strategy and vector-valued distance are very effective in taking advantages from the augmented-graph representation.

Recent years have witnessed a swelling rise of hateful and abusive content over online social networks. While detection and moderation of hate speech have been the early go-to countermeasures, the solution requires a deeper exploration of the dynamics of hate generation and propagation. We analyze more than 32 million posts from over 6.8 million users across three popular online social networks to investigate the interrelations between hateful behavior, information dissemination, and polarised organization mediated by echo chambers. We find that hatemongers play a more crucial role in governing the spread of information compared to singled-out hateful content. This observation holds for both the growth of information cascades as well as the conglomeration of hateful actors. Dissection of the core-wise distribution of these networks points towards the fact that hateful users acquire a more well-connected position in the social network and often flock together to build up information cascades. We observe that this cohesion is far from mere organized behavior; instead, in these networks, hatemongers dominate the echo chambers -- groups of users actively align themselves to specific ideological positions. The observed dominance of hateful users to inflate information cascades is primarily via user interactions amplified within these echo chambers. We conclude our study with a cautionary note that popularity-based recommendation of content is susceptible to be exploited by hatemongers given their potential to escalate content popularity via echo-chambered interactions.

The Nash equilibrium problem (NEP) [Nash, 1950, 1951] is a central topic in mathematics, economics and computer science. NEP problems have begun to play an important role in machine learning as researchers begin to focus on decisions, incentives and the dynamics of multi-agent learning. In a classical NEP, the payoff to each player depends upon the strategies chosen by all, but the domains from which the strategies are to be chosen for each player are independent of the strategies chosen by other players. The goal is to arrive at a joint optimal outcome where no player can do better by deviating unilaterally [Osborne and Rubinstein, 1994, Myerson, 2013]. The generalized Nash equilibrium problem (GNEP) is a natural generalization of an NEP where the choice of an action by one agent affects both the payoff and the domain of actions of all other agents [Arrow and Debreu, 1954]. Its introduction in the 1950's provided the foundation for a rigorous theory of economic equilibrium [Debreu, 1952, Arrow and Debreu, 1954, Debreu, 1959]. More recently, the GNEP problem has emerged as a powerful paradigm in a range of engineering applications involving noncooperative games. In particular, in the survey of Facchinei and Kanzow [2010a], three general classes of problems were developed in detail: the abstract model of general equilibrium, power allocation in a telecommunication system, and environmental pollution control.

Artificial intelligence (AI) has been making a lot of headlines lately, and with good reason. AI is quickly becoming more and more sophisticated, allowing it to be used in a variety of ways that can benefit both businesses and individuals alike. But while the potential benefits are clear, there's also some concern about how these powerful technologies may be misused or abused by those who don't have our best interests at heart. As such, many governments around the world are looking into various forms of regulation for AI technology as they seek to protect their citizens from any potentially negative consequences that could arise from its misuse. However, despite this increased focus on regulating AI technology for safety purposes – one thing remains unclear: How effective will this form of regulation really end up being?