Research Papers

Feature relevance quantification in explainable AI: A causal problem

Dominik Janzing, Lenon Minorics, Patrick Blöbaum

We discuss promising recent contributions on quantifying feature relevance using Shapley values, where we observed some confusion on which probability distribution is the right one for dropped features. We argue that the confusion is based on not carefully distinguishing between observational and interventional conditional probabilities and try a clarification based on Pearl's seminal work on causality. We conclude that unconditional rather than conditional expectations provide the right notion of dropping features in contradiction to the theoretical justification of the software package SHAP. Parts of SHAP are unaffected because unconditional expectations (which we argue to be conceptually right) are used as approximation for the conditional ones, which encouraged others to `improve' SHAP in a way that we believe to be flawed.

Should ChatGPT be Biased? Challenges and Risks of Bias in Large Language Models

Emilio Ferrara

As the capabilities of generative language models continue to advance, the implications of biases ingrained within these models have garnered increasing attention from researchers, practitioners, and the broader public. This article investigates the challenges and risks associated with biases in large-scale language models like ChatGPT. We discuss the origins of biases, stemming from, among others, the nature of training data, model specifications, algorithmic constraints, product design, and policy decisions. We explore the ethical concerns arising from the unintended consequences of biased model outputs. We further analyze the potential opportunities to mitigate biases, the inevitability of some biases, and the implications of deploying these models in various applications, such as virtual assistants, content generation, and chatbots. Finally, we review the current approaches to identify, quantify, and mitigate biases in language models, emphasizing the need for a multi-disciplinary, collaborative effort to develop more equitable, transparent, and responsible AI systems. This article aims to stimulate a thoughtful dialogue within the artificial intelligence community, encouraging researchers and developers to reflect on the role of biases in generative language models and the ongoing pursuit of ethical AI.

Measuring the Quality of Explanations: The System Causability Scale (SCS). Comparing Human and Machine Explanations

Andreas Holzinger, André Carrington, Heimo Müller

Recent success in Artificial Intelligence (AI) and Machine Learning (ML) allow problem solving automatically without any human intervention. Autonomous approaches can be very convenient. However, in certain domains, e.g., in the medical domain, it is necessary to enable a domain expert to understand, why an algorithm came up with a certain result. Consequently, the field of Explainable AI (xAI) rapidly gained interest worldwide in various domains, particularly in medicine. Explainable AI studies transparency and traceability of opaque AI/ML and there are already a huge variety of methods. For example with layer-wise relevance propagation relevant parts of inputs to, and representations in, a neural network which caused a result, can be highlighted. This is a first important step to ensure that end users, e.g., medical professionals, assume responsibility for decision making with AI/ML and of interest to professionals and regulators. Interactive ML adds the component of human expertise to AI/ML processes by enabling them to re-enact and retrace AI/ML results, e.g. let them check it for plausibility. This requires new human-AI interfaces for explainable AI. In order to build effective and efficient interactive human-AI interfaces we have to deal with the question of how to evaluate the quality of explanations given by an explainable AI system. In this paper we introduce our System Causability Scale (SCS) to measure the quality of explanations. It is based on our notion of Causability (Holzinger et al., 2019) combined with concepts adapted from a widely accepted usability scale.

Explainable AI for Trees: From Local Explanations to Global Understanding

Scott M. Lundberg, Gabriel Erion, Hugh Chen +7

Tree-based machine learning models such as random forests, decision trees, and gradient boosted trees are the most popular non-linear predictive models used in practice today, yet comparatively little attention has been paid to explaining their predictions. Here we significantly improve the interpretability of tree-based models through three main contributions: 1) The first polynomial time algorithm to compute optimal explanations based on game theory. 2) A new type of explanation that directly measures local feature interaction effects. 3) A new set of tools for understanding global model structure based on combining many local explanations of each prediction. We apply these tools to three medical machine learning problems and show how combining many high-quality local explanations allows us to represent global structure while retaining local faithfulness to the original model. These tools enable us to i) identify high magnitude but low frequency non-linear mortality risk factors in the general US population, ii) highlight distinct population sub-groups with shared risk characteristics, iii) identify non-linear interaction effects among risk factors for chronic kidney disease, and iv) monitor a machine learning model deployed in a hospital by identifying which features are degrading the model's performance over time. Given the popularity of tree-based machine learning models, these improvements to their interpretability have implications across a broad set of domains.

Low-Resource Languages Jailbreak GPT-4

Zheng-Xin Yong, Cristina Menghini, Stephen H. Bach

AI safety training and red-teaming of large language models (LLMs) are measures to mitigate the generation of unsafe content. Our work exposes the inherent cross-lingual vulnerability of these safety mechanisms, resulting from the linguistic inequality of safety training data, by successfully circumventing GPT-4's safeguard through translating unsafe English inputs into low-resource languages. On the AdvBenchmark, GPT-4 engages with the unsafe translated inputs and provides actionable items that can get the users towards their harmful goals 79% of the time, which is on par with or even surpassing state-of-the-art jailbreaking attacks. Other high-/mid-resource languages have significantly lower attack success rate, which suggests that the cross-lingual vulnerability mainly applies to low-resource languages. Previously, limited training on low-resource languages primarily affects speakers of those languages, causing technological disparities. However, our work highlights a crucial shift: this deficiency now poses a risk to all LLMs users. Publicly available translation APIs enable anyone to exploit LLMs' safety vulnerabilities. Therefore, our work calls for a more holistic red-teaming efforts to develop robust multilingual safeguards with wide language coverage.

Proxy Tasks and Subjective Measures Can Be Misleading in Evaluating Explainable AI Systems

Zana Buçinca, Phoebe Lin, Krzysztof Z. Gajos +1

Explainable artificially intelligent (XAI) systems form part of sociotechnical systems, e.g., human+AI teams tasked with making decisions. Yet, current XAI systems are rarely evaluated by measuring the performance of human+AI teams on actual decision-making tasks. We conducted two online experiments and one in-person think-aloud study to evaluate two currently common techniques for evaluating XAI systems: (1) using proxy, artificial tasks such as how well humans predict the AI's decision from the given explanations, and (2) using subjective measures of trust and preference as predictors of actual performance. The results of our experiments demonstrate that evaluations with proxy tasks did not predict the results of the evaluations with the actual decision-making tasks. Further, the subjective measures on evaluations with actual decision-making tasks did not predict the objective performance on those same tasks. Our results suggest that by employing misleading evaluation methods, our field may be inadvertently slowing its progress toward developing human+AI teams that can reliably perform better than humans or AIs alone.

Data Cards: Purposeful and Transparent Dataset Documentation for Responsible AI

Mahima Pushkarna, Andrew Zaldivar, Oddur Kjartansson

As research and industry moves towards large-scale models capable of numerous downstream tasks, the complexity of understanding multi-modal datasets that give nuance to models rapidly increases. A clear and thorough understanding of a dataset's origins, development, intent, ethical considerations and evolution becomes a necessary step for the responsible and informed deployment of models, especially those in people-facing contexts and high-risk domains. However, the burden of this understanding often falls on the intelligibility, conciseness, and comprehensiveness of the documentation. It requires consistency and comparability across the documentation of all datasets involved, and as such documentation must be treated as a user-centric product in and of itself. In this paper, we propose Data Cards for fostering transparent, purposeful and human-centered documentation of datasets within the practical contexts of industry and research. Data Cards are structured summaries of essential facts about various aspects of ML datasets needed by stakeholders across a dataset's lifecycle for responsible AI development. These summaries provide explanations of processes and rationales that shape the data and consequently the models, such as upstream sources, data collection and annotation methods; training and evaluation methods, intended use; or decisions affecting model performance. We also present frameworks that ground Data Cards in real-world utility and human-centricity. Using two case studies, we report on desirable characteristics that support adoption across domains, organizational structures, and audience groups. Finally, we present lessons learned from deploying over 20 Data Cards.

We developed an explainable artificial intelligence (AI) early warning score (xAI-EWS) system for early detection of acute critical illness. While maintaining a high predictive performance, our system explains to the clinician on which relevant electronic health records (EHRs) data the prediction is grounded. Acute critical illness is often preceded by deterioration of routinely measured clinical parameters, e.g., blood pressure and heart rate. Early clinical prediction is typically based on manually calculated screening metrics that simply weigh these parameters, such as Early Warning Scores (EWS). The predictive performance of EWSs yields a tradeoff between sensitivity and specificity that can lead to negative outcomes for the patient. Previous work on EHR-trained AI systems offers promising results with high levels of predictive performance in relation to the early, real-time prediction of acute critical illness. However, without insight into the complex decisions by such system, clinical translation is hindered. In this letter, we present our xAI-EWS system, which potentiates clinical translation by accompanying a prediction with information on the EHR data explaining it.

Auditing large language models: a three-layered approach

Jakob Mökander, Jonas Schuett, Hannah Rose Kirk +1

Large language models (LLMs) represent a major advance in artificial intelligence (AI) research. However, the widespread use of LLMs is also coupled with significant ethical and social challenges. Previous research has pointed towards auditing as a promising governance mechanism to help ensure that AI systems are designed and deployed in ways that are ethical, legal, and technically robust. However, existing auditing procedures fail to address the governance challenges posed by LLMs, which display emergent capabilities and are adaptable to a wide range of downstream tasks. In this article, we address that gap by outlining a novel blueprint for how to audit LLMs. Specifically, we propose a three-layered approach, whereby governance audits (of technology providers that design and disseminate LLMs), model audits (of LLMs after pre-training but prior to their release), and application audits (of applications based on LLMs) complement and inform each other. We show how audits, when conducted in a structured and coordinated manner on all three levels, can be a feasible and effective mechanism for identifying and managing some of the ethical and social risks posed by LLMs. However, it is important to remain realistic about what auditing can reasonably be expected to achieve. Therefore, we discuss the limitations not only of our three-layered approach but also of the prospect of auditing LLMs at all. Ultimately, this article seeks to expand the methodological toolkit available to technology providers and policymakers who wish to analyse and evaluate LLMs from technical, ethical, and legal perspectives.

Evaluating Explainable AI: Which Algorithmic Explanations Help Users Predict Model Behavior?

Peter Hase, Mohit Bansal

Algorithmic approaches to interpreting machine learning models have proliferated in recent years. We carry out human subject tests that are the first of their kind to isolate the effect of algorithmic explanations on a key aspect of model interpretability, simulatability, while avoiding important confounding experimental factors. A model is simulatable when a person can predict its behavior on new inputs. Through two kinds of simulation tests involving text and tabular data, we evaluate five explanations methods: (1) LIME, (2) Anchor, (3) Decision Boundary, (4) a Prototype model, and (5) a Composite approach that combines explanations from each method. Clear evidence of method effectiveness is found in very few cases: LIME improves simulatability in tabular classification, and our Prototype method is effective in counterfactual simulation tests. We also collect subjective ratings of explanations, but we do not find that ratings are predictive of how helpful explanations are. Our results provide the first reliable and comprehensive estimates of how explanations influence simulatability across a variety of explanation methods and data domains. We show that (1) we need to be careful about the metrics we use to evaluate explanation methods, and (2) there is significant room for improvement in current methods. All our supporting code, data, and models are publicly available at: https://github.com/peterbhase/InterpretableNLP-ACL2020

Why Language Models Hallucinate

Adam Tauman Kalai, Ofir Nachum, Santosh S. Vempala +1

Like students facing hard exam questions, large language models sometimes guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty. Such "hallucinations" persist even in state-of-the-art systems and undermine trust. We argue that language models hallucinate because the training and evaluation procedures reward guessing over acknowledging uncertainty, and we analyze the statistical causes of hallucinations in the modern training pipeline. Hallucinations need not be mysterious -- they originate simply as errors in binary classification. If incorrect statements cannot be distinguished from facts, then hallucinations in pretrained language models will arise through natural statistical pressures. We then argue that hallucinations persist due to the way most evaluations are graded -- language models are optimized to be good test-takers, and guessing when uncertain improves test performance. This "epidemic" of penalizing uncertain responses can only be addressed through a socio-technical mitigation: modifying the scoring of existing benchmarks that are misaligned but dominate leaderboards, rather than introducing additional hallucination evaluations. This change may steer the field toward more trustworthy AI systems.

This survey presents a comprehensive review of current literature on Explainable Artificial Intelligence (XAI) methods for cyber security applications. Due to the rapid development of Internet-connected systems and Artificial Intelligence in recent years, Artificial Intelligence including Machine Learning (ML) and Deep Learning (DL) has been widely utilized in the fields of cyber security including intrusion detection, malware detection, and spam filtering. However, although Artificial Intelligence-based approaches for the detection and defense of cyber attacks and threats are more advanced and efficient compared to the conventional signature-based and rule-based cyber security strategies, most ML-based techniques and DL-based techniques are deployed in the black-box manner, meaning that security experts and customers are unable to explain how such procedures reach particular conclusions. The deficiencies of transparency and interpretability of existing Artificial Intelligence techniques would decrease human users' confidence in the models utilized for the defense against cyber attacks, especially in current situations where cyber attacks become increasingly diverse and complicated. Therefore, it is essential to apply XAI in the establishment of cyber security models to create more explainable models while maintaining high accuracy and allowing human users to comprehend, trust, and manage the next generation of cyber defense mechanisms. Although there are papers reviewing Artificial Intelligence applications in cyber security areas and the vast literature on applying XAI in many fields including healthcare, financial services, and criminal justice, the surprising fact is that there are currently no survey research articles that concentrate on XAI applications in cyber security.