Research Papers

Amidst decline in public trust in technology, computing ethics have taken center stage, and critics have raised questions about corporate ethics washing. Yet few studies examine the actual implementation of AI ethics values in technology companies. Based on a qualitative analysis of technology workers tasked with integrating AI ethics into product development, we find that workers experience an environment where policies, practices, and outcomes are decoupled. We analyze AI ethics workers as ethics entrepreneurs who work to institutionalize new ethics-related practices within organizations. We show that ethics entrepreneurs face three major barriers to their work. First, they struggle to have ethics prioritized in an environment centered around software product launches. Second, ethics are difficult to quantify in a context where company goals are incentivized by metrics. Third, the frequent reorganization of teams makes it difficult to access knowledge and maintain relationships central to their work. Consequently, individuals take on great personal risk when raising ethics issues, especially when they come from marginalized backgrounds. These findings shed light on complex dynamics of institutional change at technology companies.

The AI Index 2022 Annual Report

Daniel Zhang, Nestor Maslej, Erik Brynjolfsson +10

Welcome to the fifth edition of the AI Index Report! The latest edition includes data from a broad set of academic, private, and nonprofit organizations as well as more self-collected data and original analysis than any previous editions, including an expanded technical performance chapter, a new survey of robotics researchers around the world, data on global AI legislation records in 25 countries, and a new chapter with an in-depth analysis of technical AI ethics metrics. The AI Index Report tracks, collates, distills, and visualizes data related to artificial intelligence. Its mission is to provide unbiased, rigorously vetted, and globally sourced data for policymakers, researchers, executives, journalists, and the general public to develop a more thorough and nuanced understanding of the complex field of AI. The report aims to be the world's most credible and authoritative source for data and insights about AI.

Synthetic Benchmarks for Scientific Research in Explainable Machine Learning

Yang Liu, Sujay Khandagale, Colin White +1

As machine learning models grow more complex and their applications become more high-stakes, tools for explaining model predictions have become increasingly important. This has spurred a flurry of research in model explainability and has given rise to feature attribution methods such as LIME and SHAP. Despite their widespread use, evaluating and comparing different feature attribution methods remains challenging: evaluations ideally require human studies, and empirical evaluation metrics are often data-intensive or computationally prohibitive on real-world datasets. In this work, we address this issue by releasing XAI-Bench: a suite of synthetic datasets along with a library for benchmarking feature attribution algorithms. Unlike real-world datasets, synthetic datasets allow the efficient computation of conditional expected values that are needed to evaluate ground-truth Shapley values and other metrics. The synthetic datasets we release offer a wide variety of parameters that can be configured to simulate real-world data. We demonstrate the power of our library by benchmarking popular explainability techniques across several evaluation metrics and across a variety of settings. The versatility and efficiency of our library will help researchers bring their explainability methods from development to deployment. Our code is available at https://github.com/abacusai/xai-bench.

The continuous development of artificial intelligence (AI) theory has propelled this field to unprecedented heights, owing to the relentless efforts of scholars and researchers. In the medical realm, AI takes a pivotal role, leveraging robust machine learning (ML) algorithms. AI technology in medical imaging aids physicians in X-ray, computed tomography (CT) scans, and magnetic resonance imaging (MRI) diagnoses, conducts pattern recognition and disease prediction based on acoustic data, delivers prognoses on disease types and developmental trends for patients, and employs intelligent health management wearable devices with human-computer interaction technology to name but a few. While these well-established applications have significantly assisted in medical field diagnoses, clinical decision-making, and management, collaboration between the medical and AI sectors faces an urgent challenge: How to substantiate the reliability of decision-making? The underlying issue stems from the conflict between the demand for accountability and result transparency in medical scenarios and the black-box model traits of AI. This article reviews recent research grounded in explainable artificial intelligence (XAI), with an emphasis on medical practices within the visual, audio, and multimodal perspectives. We endeavour to categorise and synthesise these practices, aiming to provide support and guidance for future researchers and healthcare professionals.

AI assistants such as ChatGPT are trained to respond to users by saying, "I am a large language model". This raises questions. Do such models know that they are LLMs and reliably act on this knowledge? Are they aware of their current circumstances, such as being deployed to the public? We refer to a model's knowledge of itself and its circumstances as situational awareness. To quantify situational awareness in LLMs, we introduce a range of behavioral tests, based on question answering and instruction following. These tests form the $\textbf{Situational Awareness Dataset (SAD)}$, a benchmark comprising 7 task categories and over 13,000 questions. The benchmark tests numerous abilities, including the capacity of LLMs to (i) recognize their own generated text, (ii) predict their own behavior, (iii) determine whether a prompt is from internal evaluation or real-world deployment, and (iv) follow instructions that depend on self-knowledge. We evaluate 16 LLMs on SAD, including both base (pretrained) and chat models. While all models perform better than chance, even the highest-scoring model (Claude 3 Opus) is far from a human baseline on certain tasks. We also observe that performance on SAD is only partially predicted by metrics of general knowledge (e.g. MMLU). Chat models, which are finetuned to serve as AI assistants, outperform their corresponding base models on SAD but not on general knowledge tasks. The purpose of SAD is to facilitate scientific understanding of situational awareness in LLMs by breaking it down into quantitative abilities. Situational awareness is important because it enhances a model's capacity for autonomous planning and action. While this has potential benefits for automation, it also introduces novel risks related to AI safety and control. Code and latest results available at https://situational-awareness-dataset.org .

Advances in machine learning have led to graph neural network-based methods for drug discovery, yielding promising results in molecular design, chemical synthesis planning, and molecular property prediction. However, current graph neural networks (GNNs) remain of limited acceptance in drug discovery is limited due to their lack of interpretability. Although this major weakness has been mitigated by the development of explainable artificial intelligence (XAI) techniques, the "ground truth" assignment in most explainable tasks ultimately rests with subjective judgments by humans so that the quality of model interpretation is hard to evaluate in quantity. In this work, we first build three levels of benchmark datasets to quantitatively assess the interpretability of the state-of-the-art GNN models. Then we implemented recent XAI methods in combination with different GNN algorithms to highlight the benefits, limitations, and future opportunities for drug discovery. As a result, GradInput and IG generally provide the best model interpretability for GNNs, especially when combined with GraphNet and CMPNN. The integrated and developed XAI package is fully open-sourced and can be used by practitioners to train new models on other drug discovery tasks.

As AI systems are integrated into high stakes social domains, researchers now examine how to design and operate them in a safe and ethical manner. However, the criteria for identifying and diagnosing safety risks in complex social contexts remain unclear and contested. In this paper, we examine the vagueness in debates about the safety and ethical behavior of AI systems. We show how this vagueness cannot be resolved through mathematical formalism alone, instead requiring deliberation about the politics of development as well as the context of deployment. Drawing from a new sociotechnical lexicon, we redefine vagueness in terms of distinct design challenges at key stages in AI system development. The resulting framework of Hard Choices in Artificial Intelligence (HCAI) empowers developers by 1) identifying points of overlap between design decisions and major sociotechnical challenges; 2) motivating the creation of stakeholder feedback channels so that safety issues can be exhaustively addressed. As such, HCAI contributes to a timely debate about the status of AI development in democratic societies, arguing that deliberation should be the goal of AI Safety, not just the procedure by which it is ensured.

AI Generated Content (AIGC) has received tremendous attention within the past few years, with content generated in the format of image, text, audio, video, etc. Meanwhile, AIGC has become a double-edged sword and recently received much criticism regarding its responsible usage. In this article, we focus on 8 main concerns that may hinder the healthy development and deployment of AIGC in practice, including risks from (1) privacy; (2) bias, toxicity, misinformation; (3) intellectual property (IP); (4) robustness; (5) open source and explanation; (6) technology abuse; (7) consent, credit, and compensation; (8) environment. Additionally, we provide insights into the promising directions for tackling these risks while constructing generative models, enabling AIGC to be used more responsibly to truly benefit society.

Contrastive Explanations with Local Foil Trees

Jasper van der Waa, Marcel Robeer, Jurriaan van Diggelen +2

Recent advances in interpretable Machine Learning (iML) and eXplainable AI (XAI) construct explanations based on the importance of features in classification tasks. However, in a high-dimensional feature space this approach may become unfeasible without restraining the set of important features. We propose to utilize the human tendency to ask questions like "Why this output (the fact) instead of that output (the foil)?" to reduce the number of features to those that play a main role in the asked contrast. Our proposed method utilizes locally trained one-versus-all decision trees to identify the disjoint set of rules that causes the tree to classify data points as the foil and not as the fact. In this study we illustrate this approach on three benchmark classification tasks.

Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress?

Richard Ren, Steven Basart, Adam Khoja +9

As artificial intelligence systems grow more powerful, there has been increasing interest in "AI safety" research to address emerging and future risks. However, the field of AI safety remains poorly defined and inconsistently measured, leading to confusion about how researchers can contribute. This lack of clarity is compounded by the unclear relationship between AI safety benchmarks and upstream general capabilities (e.g., general knowledge and reasoning). To address these issues, we conduct a comprehensive meta-analysis of AI safety benchmarks, empirically analyzing their correlation with general capabilities across dozens of models and providing a survey of existing directions in AI safety. Our findings reveal that many safety benchmarks highly correlate with both upstream model capabilities and training compute, potentially enabling "safetywashing"--where capability improvements are misrepresented as safety advancements. Based on these findings, we propose an empirical foundation for developing more meaningful safety metrics and define AI safety in a machine learning research context as a set of clearly delineated research goals that are empirically separable from generic capabilities advancements. In doing so, we aim to provide a more rigorous framework for AI safety research, advancing the science of safety evaluations and clarifying the path towards measurable progress.

Managing extreme AI risks amid rapid progress

Yoshua Bengio, Geoffrey Hinton, Andrew Yao +22

Artificial Intelligence (AI) is progressing rapidly, and companies are shifting their focus to developing generalist AI systems that can autonomously act and pursue goals. Increases in capabilities and autonomy may soon massively amplify AI's impact, with risks that include large-scale social harms, malicious uses, and an irreversible loss of human control over autonomous AI systems. Although researchers have warned of extreme risks from AI, there is a lack of consensus about how exactly such risks arise, and how to manage them. Society's response, despite promising first steps, is incommensurate with the possibility of rapid, transformative progress that is expected by many experts. AI safety research is lagging. Present governance initiatives lack the mechanisms and institutions to prevent misuse and recklessness, and barely address autonomous systems. In this short consensus paper, we describe extreme risks from upcoming, advanced AI systems. Drawing on lessons learned from other safety-critical technologies, we then outline a comprehensive plan combining technical research and development with proactive, adaptive governance mechanisms for a more commensurate preparation.

Machine learning (ML) is increasingly used to support decision-making in the healthcare sector. While ML approaches provide promising results with regard to their classification performance, most share a central limitation, namely their black-box character. Motivated by the interest to understand the functioning of ML models, methods from the field of Explainable Artificial Intelligence (XAI) have recently become important. This article investigates the usefulness of XAI methods in clinical gait classification. For this purpose, predictions of state-of-the-art classification methods are explained with an established XAI method, i.e., Layer-wise Relevance Propagation (LRP). We propose to evaluate the obtained explanations with two complementary approaches: a statistical analysis of the underlying data using Statistical Parametric Mapping and a qualitative evaluation by a clinical expert. A gait dataset comprising ground reaction force measurements from 132 patients with different lower-body gait disorders and 62 healthy controls is utilized. We investigate several gait classification tasks, employ multiple classification methods, and analyze the impact of data normalization and different signal components for classification performance and explanation quality. Our experiments show that explanations obtained by LRP exhibit promising statistical properties concerning inter-class discriminativity and are also in line with clinically relevant biomechanical gait characteristics.