Research Papers

An Adversarial Approach for Explainable AI in Intrusion Detection Systems

Daniel L. Marino, Chathurika S. Wickramasinghe, Milos Manic

Despite the growing popularity of modern machine learning techniques (e.g. Deep Neural Networks) in cyber-security applications, most of these models are perceived as a black-box for the user. Adversarial machine learning offers an approach to increase our understanding of these models. In this paper we present an approach to generate explanations for incorrect classifications made by data-driven Intrusion Detection Systems (IDSs). An adversarial approach is used to find the minimum modifications (of the input features) required to correctly classify a given set of misclassified samples. The magnitude of such modifications is used to visualize the most relevant features that explain the reason for the misclassification. The presented methodology generated satisfactory explanations that describe the reasoning behind the mis-classifications, with descriptions that match expert knowledge. The advantages of the presented methodology are: 1) applicable to any classifier with defined gradients. 2) does not require any modification of the classifier model. 3) can be extended to perform further diagnosis (e.g. vulnerability assessment) and gain further understanding of the system. Experimental evaluation was conducted on the NSL-KDD99 benchmark dataset using Linear and Multilayer perceptron classifiers. The results are shown using intuitive visualizations in order to improve the interpretability of the results.

In this multicultural age, language translation is one of the most performed tasks, and it is becoming increasingly AI-moderated and automated. As a novel AI system, ChatGPT claims to be proficient in such translation tasks and in this paper, we put that claim to the test. Specifically, we examine ChatGPT's accuracy in translating between English and languages that exclusively use gender-neutral pronouns. We center this study around Bengali, the 7$^{th}$ most spoken language globally, but also generalize our findings across five other languages: Farsi, Malay, Tagalog, Thai, and Turkish. We find that ChatGPT perpetuates gender defaults and stereotypes assigned to certain occupations (e.g. man = doctor, woman = nurse) or actions (e.g. woman = cook, man = go to work), as it converts gender-neutral pronouns in languages to `he' or `she'. We also observe ChatGPT completely failing to translate the English gender-neutral pronoun `they' into equivalent gender-neutral pronouns in other languages, as it produces translations that are incoherent and incorrect. While it does respect and provide appropriately gender-marked versions of Bengali words when prompted with gender information in English, ChatGPT appears to confer a higher respect to men than to women in the same occupation. We conclude that ChatGPT exhibits the same gender biases which have been demonstrated for tools like Google Translate or MS Translator, as we provide recommendations for a human centered approach for future designers of AIs that perform language translation to better accommodate such low-resource languages.

Responsible AI is widely considered as one of the greatest scientific challenges of our time and is key to increase the adoption of AI. Recently, a number of AI ethics principles frameworks have been published. However, without further guidance on best practices, practitioners are left with nothing much beyond truisms. Also, significant efforts have been placed at algorithm-level rather than system-level, mainly focusing on a subset of mathematics-amenable ethical principles, such as fairness. Nevertheless, ethical issues can arise at any step of the development lifecycle, cutting across many AI and non-AI components of systems beyond AI algorithms and models. To operationalize responsible AI from a system perspective, in this paper, we present a Responsible AI Pattern Catalogue based on the results of a Multivocal Literature Review (MLR). Rather than staying at the principle or algorithm level, we focus on patterns that AI system stakeholders can undertake in practice to ensure that the developed AI systems are responsible throughout the entire governance and engineering lifecycle. The Responsible AI Pattern Catalogue classifies the patterns into three groups: multi-level governance patterns, trustworthy process patterns, and responsible-AI-by-design product patterns. These patterns provide systematic and actionable guidance for stakeholders to implement responsible AI.

As the influence and use of artificial intelligence (AI) have grown and its transformative potential has become more apparent, many questions have been raised regarding the economic, political, social, and ethical implications of its use. Public opinion plays an important role in these discussions, influencing product adoption, commercial development, research funding, and regulation. In this paper we present results of an in-depth survey of public opinion of artificial intelligence conducted with 10,005 respondents spanning eight countries and six continents. We report widespread perception that AI will have significant impact on society, accompanied by strong support for the responsible development and use of AI, and also characterize the public's sentiment towards AI with four key themes (exciting, useful, worrying, and futuristic) whose prevalence distinguishes response to AI in different countries.

How can I choose an explainer? An Application-grounded Evaluation of Post-hoc Explanations

Sérgio Jesus, Catarina Belém, Vladimir Balayan +4

There have been several research works proposing new Explainable AI (XAI) methods designed to generate model explanations having specific properties, or desiderata, such as fidelity, robustness, or human-interpretability. However, explanations are seldom evaluated based on their true practical impact on decision-making tasks. Without that assessment, explanations might be chosen that, in fact, hurt the overall performance of the combined system of ML model + end-users. This study aims to bridge this gap by proposing XAI Test, an application-grounded evaluation methodology tailored to isolate the impact of providing the end-user with different levels of information. We conducted an experiment following XAI Test to evaluate three popular post-hoc explanation methods -- LIME, SHAP, and TreeInterpreter -- on a real-world fraud detection task, with real data, a deployed ML model, and fraud analysts. During the experiment, we gradually increased the information provided to the fraud analysts in three stages: Data Only, i.e., just transaction data without access to model score nor explanations, Data + ML Model Score, and Data + ML Model Score + Explanations. Using strong statistical analysis, we show that, in general, these popular explainers have a worse impact than desired. Some of the conclusion highlights include: i) showing Data Only results in the highest decision accuracy and the slowest decision time among all variants tested, ii) all the explainers improve accuracy over the Data + ML Model Score variant but still result in lower accuracy when compared with Data Only; iii) LIME was the least preferred by users, probably due to its substantially lower variability of explanations from case to case.

An Adversarial Perspective on Machine Unlearning for AI Safety

Jakub Łucki, Boyi Wei, Yangsibo Huang +3

Large language models are finetuned to refuse questions about hazardous knowledge, but these protections can often be bypassed. Unlearning methods aim at completely removing hazardous capabilities from models and make them inaccessible to adversaries. This work challenges the fundamental differences between unlearning and traditional safety post-training from an adversarial perspective. We demonstrate that existing jailbreak methods, previously reported as ineffective against unlearning, can be successful when applied carefully. Furthermore, we develop a variety of adaptive methods that recover most supposedly unlearned capabilities. For instance, we show that finetuning on 10 unrelated examples or removing specific directions in the activation space can recover most hazardous capabilities for models edited with RMU, a state-of-the-art unlearning method. Our findings challenge the robustness of current unlearning approaches and question their advantages over safety training.

Towards Guaranteed Safe AI: A Framework for Ensuring Robust and Reliable AI Systems

David "davidad" Dalrymple, Joar Skalse, Yoshua Bengio +14

Ensuring that AI systems reliably and robustly avoid harmful or dangerous behaviours is a crucial challenge, especially for AI systems with a high degree of autonomy and general intelligence, or systems used in safety-critical contexts. In this paper, we will introduce and define a family of approaches to AI safety, which we will refer to as guaranteed safe (GS) AI. The core feature of these approaches is that they aim to produce AI systems which are equipped with high-assurance quantitative safety guarantees. This is achieved by the interplay of three core components: a world model (which provides a mathematical description of how the AI system affects the outside world), a safety specification (which is a mathematical description of what effects are acceptable), and a verifier (which provides an auditable proof certificate that the AI satisfies the safety specification relative to the world model). We outline a number of approaches for creating each of these three core components, describe the main technical challenges, and suggest a number of potential solutions to them. We also argue for the necessity of this approach to AI safety, and for the inadequacy of the main alternative approaches.

Gender, Race, and Intersectional Bias in Resume Screening via Language Model Retrieval

Kyra Wilson, Aylin Caliskan

Artificial intelligence (AI) hiring tools have revolutionized resume screening, and large language models (LLMs) have the potential to do the same. However, given the biases which are embedded within LLMs, it is unclear whether they can be used in this scenario without disadvantaging groups based on their protected attributes. In this work, we investigate the possibilities of using LLMs in a resume screening setting via a document retrieval framework that simulates job candidate selection. Using that framework, we then perform a resume audit study to determine whether a selection of Massive Text Embedding (MTE) models are biased in resume screening scenarios. We simulate this for nine occupations, using a collection of over 500 publicly available resumes and 500 job descriptions. We find that the MTEs are biased, significantly favoring White-associated names in 85.1\% of cases and female-associated names in only 11.1\% of cases, with a minority of cases showing no statistically significant differences. Further analyses show that Black males are disadvantaged in up to 100\% of cases, replicating real-world patterns of bias in employment settings, and validate three hypotheses of intersectionality. We also find an impact of document length as well as the corpus frequency of names in the selection of resumes. These findings have implications for widely used AI tools that are automating employment, fairness, and tech policy.

Biased Programmers? Or Biased Data? A Field Experiment in Operationalizing AI Ethics

Bo Cowgill, Fabrizio Dell'Acqua, Samuel Deng +3

Why do biased predictions arise? What interventions can prevent them? We evaluate 8.2 million algorithmic predictions of math performance from $\approx$400 AI engineers, each of whom developed an algorithm under a randomly assigned experimental condition. Our treatment arms modified programmers' incentives, training data, awareness, and/or technical knowledge of AI ethics. We then assess out-of-sample predictions from their algorithms using randomized audit manipulations of algorithm inputs and ground-truth math performance for 20K subjects. We find that biased predictions are mostly caused by biased training data. However, one-third of the benefit of better training data comes through a novel economic mechanism: Engineers exert greater effort and are more responsive to incentives when given better training data. We also assess how performance varies with programmers' demographic characteristics, and their performance on a psychological test of implicit bias (IAT) concerning gender and careers. We find no evidence that female, minority and low-IAT engineers exhibit lower bias or discrimination in their code. However, we do find that prediction errors are correlated within demographic groups, which creates performance improvements through cross-demographic averaging. Finally, we quantify the benefits and tradeoffs of practical managerial or policy interventions such as technical advice, simple reminders, and improved incentives for decreasing algorithmic bias.

Artificial Intelligence (AI) governance regulates the exercise of authority and control over the management of AI. It aims at leveraging AI through effective use of data and minimization of AI-related cost and risk. While topics such as AI governance and AI ethics are thoroughly discussed on a theoretical, philosophical, societal and regulatory level, there is limited work on AI governance targeted to companies and corporations. This work views AI products as systems, where key functionality is delivered by machine learning (ML) models leveraging (training) data. We derive a conceptual framework by synthesizing literature on AI and related fields such as ML. Our framework decomposes AI governance into governance of data, (ML) models and (AI) systems along four dimensions. It relates to existing IT and data governance frameworks and practices. It can be adopted by practitioners and academics alike. For practitioners the synthesis of mainly research papers, but also practitioner publications and publications of regulatory bodies provides a valuable starting point to implement AI governance, while for academics the paper highlights a number of areas of AI governance that deserve more attention.

Machine learning systems have become popular in fields such as marketing, financing, or data mining. While they are highly accurate, complex machine learning systems pose challenges for engineers and users. Their inherent complexity makes it impossible to easily judge their fairness and the correctness of statistically learned relations between variables and classes. Explainable AI aims to solve this challenge by modelling explanations alongside with the classifiers, potentially improving user trust and acceptance. However, users should not be fooled by persuasive, yet untruthful explanations. We therefore conduct a user study in which we investigate the effects of model accuracy and explanation fidelity, i.e. how truthfully the explanation represents the underlying model, on user trust. Our findings show that accuracy is more important for user trust than explainability. Adding an explanation for a classification result can potentially harm trust, e.g. when adding nonsensical explanations. We also found that users cannot be tricked by high-fidelity explanations into having trust for a bad classifier. Furthermore, we found a mismatch between observed (implicit) and self-reported (explicit) trust.

Do Concept Bottleneck Models Learn as Intended?

Andrei Margeloiu, Matthew Ashman, Umang Bhatt +3

Concept bottleneck models map from raw inputs to concepts, and then from concepts to targets. Such models aim to incorporate pre-specified, high-level concepts into the learning procedure, and have been motivated to meet three desiderata: interpretability, predictability, and intervenability. However, we find that concept bottleneck models struggle to meet these goals. Using post hoc interpretability methods, we demonstrate that concepts do not correspond to anything semantically meaningful in input space, thus calling into question the usefulness of concept bottleneck models in their current form.