Research Papers

Bridging the Transparency Gap: What Can Explainable AI Learn From the AI Act?

Balint Gyevnar, Nick Ferguson, Burkhard Schafer

The European Union has proposed the Artificial Intelligence Act which introduces detailed requirements of transparency for AI systems. Many of these requirements can be addressed by the field of explainable AI (XAI), however, there is a fundamental difference between XAI and the Act regarding what transparency is. The Act views transparency as a means that supports wider values, such as accountability, human rights, and sustainable innovation. In contrast, XAI views transparency narrowly as an end in itself, focusing on explaining complex algorithmic properties without considering the socio-technical context. We call this difference the ``transparency gap''. Failing to address the transparency gap, XAI risks leaving a range of transparency issues unaddressed. To begin to bridge this gap, we overview and clarify the terminology of how XAI and European regulation -- the Act and the related General Data Protection Regulation (GDPR) -- view basic definitions of transparency. By comparing the disparate views of XAI and regulation, we arrive at four axes where practical work could bridge the transparency gap: defining the scope of transparency, clarifying the legal status of XAI, addressing issues with conformity assessment, and building explainability for datasets.

ChatGPT is not only fun to chat with, but it also searches information, answers questions, and gives advice. With consistent moral advice, it might improve the moral judgment and decisions of users, who often hold contradictory moral beliefs. Unfortunately, ChatGPT turns out highly inconsistent as a moral advisor. Nonetheless, it influences users' moral judgment, we find in an experiment, even if they know they are advised by a chatting bot, and they underestimate how much they are influenced. Thus, ChatGPT threatens to corrupt rather than improves users' judgment. These findings raise the question of how to ensure the responsible use of ChatGPT and similar AI. Transparency is often touted but seems ineffective. We propose training to improve digital literacy.

Explainable artificial intelligence (XAI) twin systems will be a fundamental enabler of zero-touch network and service management (ZSM) for sixth-generation (6G) wireless networks. A reliable XAI twin system for ZSM requires two composites: an extreme analytical ability for discretizing the physical behavior of the Internet of Everything (IoE) and rigorous methods for characterizing the reasoning of such behavior. In this paper, a novel neuro-symbolic explainable artificial intelligence twin framework is proposed to enable trustworthy ZSM for a wireless IoE. The physical space of the XAI twin executes a neural-network-driven multivariate regression to capture the time-dependent wireless IoE environment while determining unconscious decisions of IoE service aggregation. Subsequently, the virtual space of the XAI twin constructs a directed acyclic graph (DAG)-based Bayesian network that can infer a symbolic reasoning score over unconscious decisions through a first-order probabilistic language model. Furthermore, a Bayesian multi-arm bandits-based learning problem is proposed for reducing the gap between the expected explained score and the current obtained score of the proposed neuro-symbolic XAI twin. To address the challenges of extensible, modular, and stateless management functions in ZSM, the proposed neuro-symbolic XAI twin framework consists of two learning systems: 1) an implicit learner that acts as an unconscious learner in physical space, and 2) an explicit leaner that can exploit symbolic reasoning based on implicit learner decisions and prior evidence. Experimental results show that the proposed neuro-symbolic XAI twin can achieve around 96.26% accuracy while guaranteeing from 18% to 44% more trust score in terms of reasoning and closed-loop automation.

Best Practices in the Creation and Use of Emotion Lexicons

Saif M. Mohammad

Words play a central role in how we express ourselves. Lexicons of word-emotion associations are widely used in research and real-world applications for sentiment analysis, tracking emotions associated with products and policies, studying health disorders, tracking emotional arcs of stories, and so on. However, inappropriate and incorrect use of these lexicons can lead to not just sub-optimal results, but also inferences that are directly harmful to people. This paper brings together ideas from Affective Computing and AI Ethics to present, some of the practical and ethical considerations involved in the creation and use of emotion lexicons -- best practices. The goal is to provide a comprehensive set of relevant considerations, so that readers (especially those new to work with emotions) can find relevant information in one place. We hope this work will facilitate more thoughtfulness when one is deciding on what emotions to work on, how to create an emotion lexicon, how to use an emotion lexicon, how to draw meaningful inferences, and how to judge success.

As data-driven systems are increasingly deployed at scale, ethical concerns have arisen around unfair and discriminatory outcomes for historically marginalized groups that are underrepresented in training data. In response, work around AI fairness and inclusion has called for datasets that are representative of various demographic groups. In this paper, we contribute an analysis of the representativeness of age, gender, and race & ethnicity in accessibility datasets - datasets sourced from people with disabilities and older adults - that can potentially play an important role in mitigating bias for inclusive AI-infused applications. We examine the current state of representation within datasets sourced by people with disabilities by reviewing publicly-available information of 190 datasets, we call these accessibility datasets. We find that accessibility datasets represent diverse ages, but have gender and race representation gaps. Additionally, we investigate how the sensitive and complex nature of demographic variables makes classification difficult and inconsistent (e.g., gender, race & ethnicity), with the source of labeling often unknown. By reflecting on the current challenges and opportunities for representation of disabled data contributors, we hope our effort expands the space of possibility for greater inclusion of marginalized communities in AI-infused systems.

Within the current AI ethics discourse, there is a gap in empirical research on understanding how AI practitioners understand ethics and socially organize to operationalize ethical concerns, particularly in the context of AI start-ups. This gap intensifies the risk of a disconnect between scholarly research, innovation, and application. This risk materializes acutely as mounting pressures to identify and mitigate the potential harms of AI systems have created an urgent need to assess and implement socio-technical innovation for fairness, accountability, and transparency. Building on social practice theory, we address this need via a framework that allows AI researchers, practitioners, and regulators to systematically analyze existing cultural understandings, histories, and social practices of ethical AI to define appropriate strategies for effectively implementing socio-technical innovations. Our contributions are threefold: 1) we introduce a practice-based approach for understanding ethical AI; 2) we present empirical findings from our study on the operationalization of ethics in German AI start-ups to underline that AI ethics and social practices must be understood in their specific cultural and historical contexts; and 3) based on our empirical findings, we suggest that ethical AI practices can be broken down into principles, needs, narratives, materializations, and cultural genealogies to form a useful backdrop for considering socio-technical innovations.

Landslides have been a regular occurrence and an alarming threat to human life and property in the era of anthropogenic global warming. An early prediction of landslide susceptibility using a data-driven approach is a demand of time. In this study, we explored the eloquent features that best describe landslide susceptibility with state-of-the-art machine learning methods. In our study, we employed state-of-the-art machine learning algorithms including XgBoost, LR, KNN, SVM, and Adaboost for landslide susceptibility prediction. To find the best hyperparameters of each individual classifier for optimized performance, we have incorporated the Grid Search method, with 10 Fold Cross-Validation. In this context, the optimized version of XgBoost outperformed all other classifiers with a Cross-validation Weighted F1 score of 94.62 %. Followed by this empirical evidence, we explored the XgBoost classifier by incorporating TreeSHAP, a game-theory-based statistical algorithm used to explain Machine Learning models, to identify eloquent features such as SLOPE, ELEVATION, TWI that complement the performance of the XGBoost classifier mostly and features such as LANDUSE, NDVI, SPI which has less effect on models performance. According to the TreeSHAP explanation of features, we selected the 9 most significant landslide causal factors out of 15. Evidently, an optimized version of XgBoost along with feature reduction by 40 % has outperformed all other classifiers in terms of popular evaluation metrics with a Cross-Validation Weighted F1 score of 95.01 % on the training and AUC score of 97 %

Scrutinizing XAI using linear ground-truth data with suppressor variables

Rick Wilming, Céline Budding, Klaus-Robert Müller +1

Machine learning (ML) is increasingly often used to inform high-stakes decisions. As complex ML models (e.g., deep neural networks) are often considered black boxes, a wealth of procedures has been developed to shed light on their inner workings and the ways in which their predictions come about, defining the field of 'explainable AI' (XAI). Saliency methods rank input features according to some measure of 'importance'. Such methods are difficult to validate since a formal definition of feature importance is, thus far, lacking. It has been demonstrated that some saliency methods can highlight features that have no statistical association with the prediction target (suppressor variables). To avoid misinterpretations due to such behavior, we propose the actual presence of such an association as a necessary condition and objective preliminary definition for feature importance. We carefully crafted a ground-truth dataset in which all statistical dependencies are well-defined and linear, serving as a benchmark to study the problem of suppressor variables. We evaluate common explanation methods including LRP, DTD, PatternNet, PatternAttribution, LIME, Anchors, SHAP, and permutation-based methods with respect to our objective definition. We show that most of these methods are unable to distinguish important features from suppressors in this setting.

Ethics statements have been proposed as a mechanism to increase transparency and promote reflection on the societal impacts of published research. In 2020, the machine learning (ML) conference NeurIPS broke new ground by requiring that all papers include a broader impact statement. This requirement was removed in 2021, in favour of a checklist approach. The 2020 statements therefore provide a unique opportunity to learn from the broader impact experiment: to investigate the benefits and challenges of this and similar governance mechanisms, as well as providing an insight into how ML researchers think about the societal impacts of their own work. Such learning is needed as NeurIPS and other venues continue to question and adapt their policies. To enable this, we have created a dataset containing the impact statements from all NeurIPS 2020 papers, along with additional information such as affiliation type, location and subject area, and a simple visualisation tool for exploration. We also provide an initial quantitative analysis of the dataset, covering representation, engagement, common themes, and willingness to discuss potential harms alongside benefits. We investigate how these vary by geography, affiliation type and subject area. Drawing on these findings, we discuss the potential benefits and negative outcomes of ethics statement requirements, and their possible causes and associated challenges. These lead us to several lessons to be learnt from the 2020 requirement: (i) the importance of creating the right incentives, (ii) the need for clear expectations and guidance, and (iii) the importance of transparency and constructive deliberation. We encourage other researchers to use our dataset to provide additional analysis, to further our understanding of how researchers responded to this requirement, and to investigate the benefits and challenges of this and related mechanisms.

We study whether receiving advice from either a human or algorithmic advisor, accompanied by five types of Local and Global explanation labelings, has an effect on the readiness to adopt, willingness to pay, and trust in a financial AI consultant. We compare the differences over time and in various key situations using a unique experimental framework where participants play a web-based game with real monetary consequences. We observed that accuracy-based explanations of the model in initial phases leads to higher adoption rates. When the performance of the model is immaculate, there is less importance associated with the kind of explanation for adoption. Using more elaborate feature-based or accuracy-based explanations helps substantially in reducing the adoption drop upon model failure. Furthermore, using an autopilot increases adoption significantly. Participants assigned to the AI-labeled advice with explanations were willing to pay more for the advice than the AI-labeled advice with a No-explanation alternative. These results add to the literature on the importance of XAI for algorithmic adoption and trust.

This article philosophically analyzes online exam supervision technologies, which have been thrust into the public spotlight due to campus lockdowns during the COVID-19 pandemic and the growing demand for online courses. Online exam proctoring technologies purport to provide effective oversight of students sitting online exams, using artificial intelligence (AI) systems and human invigilators to supplement and review those systems. Such technologies have alarmed some students who see them as `Big Brother-like', yet some universities defend their judicious use. Critical ethical appraisal of online proctoring technologies is overdue. This article philosophically analyzes these technologies, focusing on the ethical concepts of academic integrity, fairness, non-maleficence, transparency, privacy, respect for autonomy, liberty, and trust. Most of these concepts are prominent in the new field of AI ethics and all are relevant to the education context. The essay provides ethical considerations that educational institutions will need to carefully review before electing to deploy and govern specific online proctoring technologies.

Natural Language Generation Challenges for Explainable AI

Ehud Reiter

Good quality explanations of artificial intelligence (XAI) reasoning must be written (and evaluated) for an explanatory purpose, targeted towards their readers, have a good narrative and causal structure, and highlight where uncertainty and data quality affect the AI output. I discuss these challenges from a Natural Language Generation (NLG) perspective, and highlight four specific NLG for XAI research challenges.