Information retrieval, specifically contract clause retrieval, is foundational to contract drafting because lawyers rarely draft contracts from scratch; instead, they locate and revise the most relevant precedent. We introduce the Atticus Clause Retrieval Dataset (ACORD), the first retrieval benchmark for contract drafting fully annotated by experts. ACORD focuses on complex contract clauses such as Limitation of Liability, Indemnification, Change of Control, and Most Favored Nation. It includes 114 queries and over 126,000 query-clause pairs, each ranked on a scale from 1 to 5 stars. The task is to find the most relevant precedent clauses to a query. The bi-encoder retriever paired with pointwise LLMs re-rankers shows promising results. However, substantial improvements are still needed to effectively manage the complex legal work typically undertaken by lawyers. As the first retrieval benchmark for contract drafting annotated by experts, ACORD can serve as a valuable IR benchmark for the NLP community.

One of the most promising use-cases for machine learning in industrial manufacturing is the early detection of defective products using a quality control system. Such a system can save costs and reduces human errors due to the monotonous nature of visual inspections. Today, a rich body of research exists which employs machine learning methods to identify rare defective products in unbalanced visual quality control datasets. These methods typically rely on two components: A visual backbone to capture the features of the input image and an anomaly detection algorithm that decides if these features are within an expected distribution. With the rise of transformer architecture as visual backbones of choice, there exists now a great variety of different combinations of these two components, ranging all along the trade-off between detection quality and inference time. Facing this variety, practitioners in the field often have to spend a considerable amount of time on researching the right combination for their use-case at hand. Our contribution is to help practitioners with this choice by reviewing and evaluating current vision transformer models together with anomaly detection methods. For this, we chose SotA models of both disciplines, combined them and evaluated them towards the goal of having small, fast and efficient anomaly detection models suitable for industrial manufacturing. We evaluated the results of our experiments on the well-known MVTecAD and BTAD datasets. Moreover, we give guidelines for choosing a suitable model architecture for a quality control system in practice, considering given use-case and hardware constraints.

Artificial Intelligence is rapidly embedding itself within militaries, economies, and societies, reshaping their very foundations. Given the depth and breadth of its consequences, it has never been more pressing to understand how to ensure that AI systems are safe, ethical, and have a positive societal impact. This book aims to provide a comprehensive approach to understanding AI risk. Our primary goals include consolidating fragmented knowledge on AI risk, increasing the precision of core ideas, and reducing barriers to entry by making content simpler and more comprehensible. The book has been designed to be accessible to readers from diverse backgrounds. You do not need to have studied AI, philosophy, or other such topics. The content is skimmable and somewhat modular, so that you can choose which chapters to read. We introduce mathematical formulas in a few places to specify claims more precisely, but readers should be able to understand the main points without these.

This paper introduces Ali-AUG, a novel single-step diffusion model for efficient labeled data augmentation in industrial applications. Our method addresses the challenge of limited labeled data by generating synthetic, labeled images with precise feature insertion. Ali-AUG utilizes a stable diffusion architecture enhanced with skip connections and LoRA modules to efficiently integrate masks and images, ensuring accurate feature placement without affecting unrelated image content. Experimental validation across various industrial datasets demonstrates Ali-AUG's superiority in generating high-quality, defect-enhanced images while maintaining rapid single-step inference. By offering precise control over feature insertion and minimizing required training steps, our technique significantly enhances data augmentation capabilities, providing a powerful tool for improving the performance of deep learning models in scenarios with limited labeled data. Ali-AUG is especially useful for use cases like defective product image generation to train AI-based models to improve their ability to detect defects in manufacturing processes. Using different data preparation strategies, including Classification Accuracy Score (CAS) and Naive Augmentation Score (NAS), we show that Ali-AUG improves model performance by 31% compared to other augmentation methods and by 45% compared to models without data augmentation. Notably, Ali-AUG reduces training time by 32% and supports both paired and unpaired datasets, enhancing flexibility in data preparation.

Anomaly detection (AD) plays a pivotal role in multimedia applications for detecting defective products and automating quality inspection. Deep learning (DL) models typically require large-scale annotated data, which are often highly imbalanced since anomalies are usually scarce. The black box nature of these models prohibits them from being trusted by users. To address these challenges, we propose MeLIAD, a novel methodology for interpretable anomaly detection, which unlike the previous methods is based on metric learning and achieves interpretability by design without relying on any prior distribution assumptions of true anomalies. MeLIAD requires only a few samples of anomalies for training, without employing any augmentation techniques, and is inherently interpretable, providing visualizations that offer insights into why an image is identified as anomalous. This is achieved by introducing a novel trainable entropy-based scoring component for the identification and localization of anomalous instances, and a novel loss function that jointly optimizes the anomaly scoring component with a metric learning objective. Experiments on five public benchmark datasets, including quantitative and qualitative evaluation of interpretability, demonstrate that MeLIAD achieves improved anomaly detection and localization performance compared to state-of-the-art methods.

As generative Artificial Intelligence (genAI) technologies proliferate across sectors, they offer significant benefits but also risk exacerbating discrimination. This chapter explores how genAI intersects with non-discrimination laws, identifying shortcomings and suggesting improvements. It highlights two main types of discriminatory outputs: (i) demeaning and abusive content and (ii) subtler biases due to inadequate representation of protected groups, which may not be overtly discriminatory in individual cases but have cumulative discriminatory effects. For example, genAI systems may predominantly depict white men when asked for images of people in important jobs. This chapter examines these issues, categorizing problematic outputs into three legal categories: discriminatory content; harassment; and legally hard cases like unbalanced content, harmful stereotypes or misclassification. It argues for holding genAI providers and deployers liable for discriminatory outputs and highlights the inadequacy of traditional legal frameworks to address genAI-specific issues. The chapter suggests updating EU laws, including the AI Act, to mitigate biases in training and input data, mandating testing and auditing, and evolving legislation to enforce standards for bias mitigation and inclusivity as technology advances.

In the context of industrially mass-manufactured products, quality management is based on physically inspecting a small sample from a large batch and reasoning about the batch's quality conformance. When complementing physical inspections with predictions from machine learning models, it is crucial that the uncertainty of the prediction is known. Otherwise, the application of established quality management concepts is not legitimate. Deterministic (machine learning) models lack quantification of their predictive uncertainty and are therefore unsuitable. Probabilistic (machine learning) models provide a predictive uncertainty along with the prediction. However, a concise relationship is missing between the measurement uncertainty of physical inspections and the predictive uncertainty of probabilistic models in their application in quality management. Here, we show how the predictive uncertainty of probabilistic (machine learning) models is related to the measurement uncertainty of physical inspections. This enables the use of probabilistic models for virtual inspections and integrates them into existing quality management concepts. Thus, we can provide a virtual measurement for any quality characteristic based on the process data and achieve a 100 percent inspection rate. In the field of Predictive Quality, the virtual measurement is of great interest. Based on our results, physical inspections with a low sampling rate can be accompanied by virtual measurements that allow an inspection rate of 100 percent. We add substantial value, especially to complex process chains, as faulty products/parts are identified promptly and upcoming process steps can be aborted.

Automated detection of defects in Printed Circuit Board (PCB) manufacturing using Solder Paste Inspection (SPI) and Automated Optical Inspection (AOI) machines can help improve operational efficiency and significantly reduce the need for manual intervention. In this paper, using SPI-extracted features of 6 million pins, we demonstrate a data-centric approach to train Machine Learning (ML) models to detect PCB defects at three stages of PCB manufacturing. The 6 million PCB pins correspond to 2 million components that belong to 15,387 PCBs. Using a base extreme gradient boosting (XGBoost) ML model, we iterate on the data pre-processing step to improve detection performance. Combining pin-level SPI features using component and PCB IDs, we developed training instances also at the component and PCB level. This allows the ML model to capture any inter-pin, inter-component, or spatial effects that may not be apparent at the pin level. Models are trained at the pin, component, and PCB levels, and the detection results from the different models are combined to identify defective components.

Our interdisciplinary study investigates how effectively U.S. laws confront the challenges posed by Generative AI to human values. Through an analysis of diverse hypothetical scenarios crafted during an expert workshop, we have identified notable gaps and uncertainties within the existing legal framework regarding the protection of fundamental values, such as privacy, autonomy, dignity, diversity, equity, and physical/mental well-being. Constitutional and civil rights, it appears, may not provide sufficient protection against AI-generated discriminatory outputs. Furthermore, even if we exclude the liability shield provided by Section 230, proving causation for defamation and product liability claims is a challenging endeavor due to the intricate and opaque nature of AI systems. To address the unique and unforeseeable threats posed by Generative AI, we advocate for legal frameworks that evolve to recognize new threats and provide proactive, auditable guidelines to industry stakeholders. Addressing these issues requires deep interdisciplinary collaborations to identify harms, values, and mitigation strategies.

Generative AI, in particular text-based "foundation models" (large models trained on a huge variety of information including the internet), can generate speech that could be problematic under a wide range of liability regimes. Machine learning practitioners regularly "red team" models to identify and mitigate such problematic speech: from "hallucinations" falsely accusing people of serious misconduct to recipes for constructing an atomic bomb. A key question is whether these red-teamed behaviors actually present any liability risk for model creators and deployers under U.S. law, incentivizing investments in safety mechanisms. We examine three liability regimes, tying them to common examples of red-teamed model behaviors: defamation, speech integral to criminal conduct, and wrongful death. We find that any Section 230 immunity analysis or downstream liability analysis is intimately wrapped up in the technical details of algorithm design. And there are many roadblocks to truly finding models (and their associated parties) liable for generated speech. We argue that AI should not be categorically immune from liability in these scenarios and that as courts grapple with the already fine-grained complexities of platform algorithms, the technical details of generative AI loom above with thornier questions. Courts and policymakers should think carefully about what technical design incentives they create as they evaluate these issues.