Research Papers

Contrastive Explanations for Reinforcement Learning in terms of Expected Consequences

Jasper van der Waa, Jurriaan van Diggelen, Karel van den Bosch +1

Machine Learning models become increasingly proficient in complex tasks. However, even for experts in the field, it can be difficult to understand what the model learned. This hampers trust and acceptance, and it obstructs the possibility to correct the model. There is therefore a need for transparency of machine learning models. The development of transparent classification models has received much attention, but there are few developments for achieving transparent Reinforcement Learning (RL) models. In this study we propose a method that enables a RL agent to explain its behavior in terms of the expected consequences of state transitions and outcomes. First, we define a translation of states and actions to a description that is easier to understand for human users. Second, we developed a procedure that enables the agent to obtain the consequences of a single action, as well as its entire policy. The method calculates contrasts between the consequences of a policy derived from a user query, and of the learned policy of the agent. Third, a format for generating explanations was constructed. A pilot survey study was conducted to explore preferences of users for different explanation properties. Results indicate that human users tend to favor explanations about policy rather than about single actions.

Towards Human-AI Deliberation: Design and Evaluation of LLM-Empowered Deliberative AI for AI-Assisted Decision-Making

Shuai Ma, Qiaoyi Chen, Xinru Wang +4

In AI-assisted decision-making, humans often passively review AI's suggestion and decide whether to accept or reject it as a whole. In such a paradigm, humans are found to rarely trigger analytical thinking and face difficulties in communicating the nuances of conflicting opinions to the AI when disagreements occur. To tackle this challenge, we propose Human-AI Deliberation, a novel framework to promote human reflection and discussion on conflicting human-AI opinions in decision-making. Based on theories in human deliberation, this framework engages humans and AI in dimension-level opinion elicitation, deliberative discussion, and decision updates. To empower AI with deliberative capabilities, we designed Deliberative AI, which leverages large language models (LLMs) as a bridge between humans and domain-specific models to enable flexible conversational interactions and faithful information provision. An exploratory evaluation on a graduate admissions task shows that Deliberative AI outperforms conventional explainable AI (XAI) assistants in improving humans' appropriate reliance and task performance. Based on a mixed-methods analysis of participant behavior, perception, user experience, and open-ended feedback, we draw implications for future AI-assisted decision tool design.

Charting the Sociotechnical Gap in Explainable AI: A Framework to Address the Gap in XAI

Upol Ehsan, Koustuv Saha, Munmun De Choudhury +1

Explainable AI (XAI) systems are sociotechnical in nature; thus, they are subject to the sociotechnical gap--divide between the technical affordances and the social needs. However, charting this gap is challenging. In the context of XAI, we argue that charting the gap improves our problem understanding, which can reflexively provide actionable insights to improve explainability. Utilizing two case studies in distinct domains, we empirically derive a framework that facilitates systematic charting of the sociotechnical gap by connecting AI guidelines in the context of XAI and elucidating how to use them to address the gap. We apply the framework to a third case in a new domain, showcasing its affordances. Finally, we discuss conceptual implications of the framework, share practical considerations in its operationalization, and offer guidance on transferring it to new contexts. By making conceptual and practical contributions to understanding the sociotechnical gap in XAI, the framework expands the XAI design space.

Predictive modeling is increasingly being employed to assist human decision-makers. One purported advantage of replacing human judgment with computer models in high stakes settings-- such as sentencing, hiring, policing, college admissions, and parole decisions-- is the perceived "neutrality" of computers. It is argued that because computer models do not hold personal prejudice, the predictions they produce will be equally free from prejudice. There is growing recognition that employing algorithms does not remove the potential for bias, and can even amplify it, since training data were inevitably generated by a process that is itself biased. In this paper, we provide a probabilistic definition of algorithmic bias. We propose a method to remove bias from predictive models by removing all information regarding protected variables from the permitted training data. Unlike previous work in this area, our framework is general enough to accommodate arbitrary data types, e.g. binary, continuous, etc. Motivated by models currently in use in the criminal justice system that inform decisions on pre-trial release and paroling, we apply our proposed method to a dataset on the criminal histories of individuals at the time of sentencing to produce "race-neutral" predictions of re-arrest. In the process, we demonstrate that the most common approach to creating "race-neutral" models-- omitting race as a covariate-- still results in racially disparate predictions. We then demonstrate that the application of our proposed method to these data removes racial disparities from predictions with minimal impact on predictive accuracy.

HuntGPT: Integrating Machine Learning-Based Anomaly Detection and Explainable AI with Large Language Models (LLMs)

Tarek Ali, Panos Kostakos

Machine learning (ML) is crucial in network anomaly detection for proactive threat hunting, reducing detection and response times significantly. However, challenges in model training, maintenance, and frequent false positives impact its acceptance and reliability. Explainable AI (XAI) attempts to mitigate these issues, allowing cybersecurity teams to assess AI-generated alerts with confidence, but has seen limited acceptance from incident responders. Large Language Models (LLMs) present a solution through discerning patterns in extensive information and adapting to different functional requirements. We present HuntGPT, a specialized intrusion detection dashboard applying a Random Forest classifier using the KDD99 dataset, integrating XAI frameworks like SHAP and Lime for user-friendly and intuitive model interaction, and combined with a GPT-3.5 Turbo, it delivers threats in an understandable format. The paper delves into the system's architecture, components, and technical accuracy, assessed through Certified Information Security Manager (CISM) Practice Exams, evaluating response quality across six metrics. The results demonstrate that conversational agents, supported by LLM and integrated with XAI, provide robust, explainable, and actionable AI solutions in intrusion detection, enhancing user understanding and interactive experience.

Supporting Human-AI Collaboration in Auditing LLMs with LLMs

Charvi Rastogi, Marco Tulio Ribeiro, Nicholas King +2

Large language models are becoming increasingly pervasive and ubiquitous in society via deployment in sociotechnical systems. Yet these language models, be it for classification or generation, have been shown to be biased and behave irresponsibly, causing harm to people at scale. It is crucial to audit these language models rigorously. Existing auditing tools leverage either or both humans and AI to find failures. In this work, we draw upon literature in human-AI collaboration and sensemaking, and conduct interviews with research experts in safe and fair AI, to build upon the auditing tool: AdaTest (Ribeiro and Lundberg, 2022), which is powered by a generative large language model (LLM). Through the design process we highlight the importance of sensemaking and human-AI communication to leverage complementary strengths of humans and generative models in collaborative auditing. To evaluate the effectiveness of the augmented tool, AdaTest++, we conduct user studies with participants auditing two commercial language models: OpenAI's GPT-3 and Azure's sentiment analysis model. Qualitative analysis shows that AdaTest++ effectively leverages human strengths such as schematization, hypothesis formation and testing. Further, with our tool, participants identified a variety of failures modes, covering 26 different topics over 2 tasks, that have been shown before in formal audits and also those previously under-reported.

The safety-critical scenarios of artificial intelligence (AI), such as autonomous driving, Internet of Things, smart healthcare, etc., have raised critical requirements of trustworthy AI to guarantee the privacy and security with reliable decisions. As a nascent branch for trustworthy AI, federated learning (FL) has been regarded as a promising privacy preserving framework for training a global AI model over collaborative devices. However, security challenges still exist in the FL framework, e.g., Byzantine attacks from malicious devices, and model tampering attacks from malicious server, which will degrade or destroy the accuracy of trained global AI model. In this paper, we shall propose a decentralized blockchain based FL (B-FL) architecture by using a secure global aggregation algorithm to resist malicious devices, and deploying practical Byzantine fault tolerance consensus protocol with high effectiveness and low energy consumption among multiple edge servers to prevent model tampering from the malicious server. However, to implement B-FL system at the network edge, multiple rounds of cross-validation in blockchain consensus protocol will induce long training latency. We thus formulate a network optimization problem that jointly considers bandwidth and power allocation for the minimization of long-term average training latency consisting of progressive learning rounds. We further propose to transform the network optimization problem as a Markov decision process and leverage the deep reinforcement learning based algorithm to provide high system performance with low computational complexity. Simulation results demonstrate that B-FL can resist malicious attacks from edge devices and servers, and the training latency of B-FL can be significantly reduced by deep reinforcement learning based algorithm compared with baseline algorithms.

Deep learning has the potential to automate many clinically useful tasks in medical imaging. However translation of deep learning into clinical practice has been hindered by issues such as lack of the transparency and interpretability in these "black box" algorithms compared to traditional statistical methods. Specifically, many clinical deep learning models lack rigorous and robust techniques for conveying certainty (or lack thereof) in their predictions -- ultimately limiting their appeal for extensive use in medical decision-making. Furthermore, numerous demonstrations of algorithmic bias have increased hesitancy towards deployment of deep learning for clinical applications. To this end, we explore how conformal predictions can complement existing deep learning approaches by providing an intuitive way of expressing uncertainty while facilitating greater transparency to clinical users. In this paper, we conduct field interviews with radiologists to assess possible use-cases for conformal predictors. Using insights gathered from these interviews, we devise two clinical use-cases and empirically evaluate several methods of conformal predictions on a dermatology photography dataset for skin lesion classification. We show how to modify conformal predictions to be more adaptive to subgroup differences in patient skin tones through equalized coverage. Finally, we compare conformal prediction against measures of epistemic uncertainty.

How Robust is GPT-3.5 to Predecessors? A Comprehensive Study on Language Understanding Tasks

Xuanting Chen, Junjie Ye, Can Zu +7

The GPT-3.5 models have demonstrated impressive performance in various Natural Language Processing (NLP) tasks, showcasing their strong understanding and reasoning capabilities. However, their robustness and abilities to handle various complexities of the open world have yet to be explored, which is especially crucial in assessing the stability of models and is a key aspect of trustworthy AI. In this study, we perform a comprehensive experimental analysis of GPT-3.5, exploring its robustness using 21 datasets (about 116K test samples) with 66 text transformations from TextFlint that cover 9 popular Natural Language Understanding (NLU) tasks. Our findings indicate that while GPT-3.5 outperforms existing fine-tuned models on some tasks, it still encounters significant robustness degradation, such as its average performance dropping by up to 35.74\% and 43.59\% in natural language inference and sentiment analysis tasks, respectively. We also show that GPT-3.5 faces some specific robustness challenges, including robustness instability, prompt sensitivity, and number sensitivity. These insights are valuable for understanding its limitations and guiding future research in addressing these challenges to enhance GPT-3.5's overall performance and generalization abilities.

Explainable AI improves task performance in human-AI collaboration

Julian Senoner, Simon Schallmoser, Bernhard Kratzwald +2

Artificial intelligence (AI) provides considerable opportunities to assist human work. However, one crucial challenge of human-AI collaboration is that many AI algorithms operate in a black-box manner where the way how the AI makes predictions remains opaque. This makes it difficult for humans to validate a prediction made by AI against their own domain knowledge. For this reason, we hypothesize that augmenting humans with explainable AI as a decision aid improves task performance in human-AI collaboration. To test this hypothesis, we analyze the effect of augmenting domain experts with explainable AI in the form of visual heatmaps. We then compare participants that were either supported by (a) black-box AI or (b) explainable AI, where the latter supports them to follow AI predictions when the AI is accurate or overrule the AI when the AI predictions are wrong. We conducted two preregistered experiments with representative, real-world visual inspection tasks from manufacturing and medicine. The first experiment was conducted with factory workers from an electronics factory, who performed $N=9,600$ assessments of whether electronic products have defects. The second experiment was conducted with radiologists, who performed $N=5,650$ assessments of chest X-ray images to identify lung lesions. The results of our experiments with domain experts performing real-world tasks show that task performance improves when participants are supported by explainable AI instead of black-box AI. For example, in the manufacturing setting, we find that augmenting participants with explainable AI (as opposed to black-box AI) leads to a five-fold decrease in the median error rate of human decisions, which gives a significant improvement in task performance.

Despite the widespread use of artificial intelligence (AI), designing user experiences (UX) for AI-powered systems remains challenging. UX designers face hurdles understanding AI technologies, such as pre-trained language models, as design materials. This limits their ability to ideate and make decisions about whether, where, and how to use AI. To address this problem, we bridge the literature on AI design and AI transparency to explore whether and how frameworks for transparent model reporting can support design ideation with pre-trained models. By interviewing 23 UX practitioners, we find that practitioners frequently work with pre-trained models, but lack support for UX-led ideation. Through a scenario-based design task, we identify common goals that designers seek model understanding for and pinpoint their model transparency information needs. Our study highlights the pivotal role that UX designers can play in Responsible AI and calls for supporting their understanding of AI limitations through model transparency and interrogation.

How can humans remain in control of artificial intelligence (AI)-based systems designed to perform tasks autonomously? Such systems are increasingly ubiquitous, creating benefits - but also undesirable situations where moral responsibility for their actions cannot be properly attributed to any particular person or group. The concept of meaningful human control has been proposed to address responsibility gaps and mitigate them by establishing conditions that enable a proper attribution of responsibility for humans; however, clear requirements for researchers, designers, and engineers are yet inexistent, making the development of AI-based systems that remain under meaningful human control challenging. In this paper, we address the gap between philosophical theory and engineering practice by identifying, through an iterative process of abductive thinking, four actionable properties for AI-based systems under meaningful human control, which we discuss making use of two applications scenarios: automated vehicles and AI-based hiring. First, a system in which humans and AI algorithms interact should have an explicitly defined domain of morally loaded situations within which the system ought to operate. Second, humans and AI agents within the system should have appropriate and mutually compatible representations. Third, responsibility attributed to a human should be commensurate with that human's ability and authority to control the system. Fourth, there should be explicit links between the actions of the AI agents and actions of humans who are aware of their moral responsibility. We argue that these four properties will support practically-minded professionals to take concrete steps toward designing and engineering for AI systems that facilitate meaningful human control.