The recent advances of deep learning have dramatically changed how machine learning, especially in the domain of natural language processing, can be applied to legal domain. However, this shift to the data-driven approaches calls for larger and more diverse datasets, which are nevertheless still small in number, especially in non-English languages. Here we present the first large-scale benchmark of Korean legal AI datasets, LBOX OPEN, that consists of one legal corpus, two classification tasks, two legal judgement prediction (LJP) tasks, and one summarization task. The legal corpus consists of 147k Korean precedents (259M tokens), of which 63k are sentenced in last 4 years and 96k are from the first and the second level courts in which factual issues are reviewed. The two classification tasks are case names (11.3k) and statutes (2.8k) prediction from the factual description of individual cases. The LJP tasks consist of (1) 10.5k criminal examples where the model is asked to predict fine amount, imprisonment with labor, and imprisonment without labor ranges for the given facts, and (2) 4.7k civil examples where the inputs are facts and claim for relief and outputs are the degrees of claim acceptance. The summarization task consists of the Supreme Court precedents and the corresponding summaries (20k). We also release realistic variants of the datasets by extending the domain (1) to infrequent case categories in case name (31k examples) and statute (17.7k) classification tasks, and (2) to long input sequences in the summarization task (51k). Finally, we release LCUBE, the first Korean legal language model trained on the legal corpus from this study. Given the uniqueness of the Law of South Korea and the diversity of the legal tasks covered in this work, we believe that LBOX OPEN contributes to the multilinguality of global legal research. LBOX OPEN and LCUBE will be publicly available.

Federated learning is a newly emerging paradigm for machine learning where multiple data holders (clients) collaborate to train a model on their combined dataset. Clients only share partially trained models and other statistics computed from their dataset, keeping their raw data local and private [McMahan et al., 2017, Kairouz et al., 2021]. By obviating the need for a third party to collect and store clients' data, federated learning has several advantages over the classical, centralized paradigm [Dean et al., 2012, Iandola et al., 2016, Goyal et al., 2017]: it ensures clients' consent is tied to the specific task at hand by requiring active participation of the clients in training, confers some basic level of privacy, and has the potential to make machine learning more participatory in general [Kulynych et al., 2020, Jones and Tonetti, 2020]. Further, widespread legislation of data portability and privacy requirements (such as GDPR and CCPA) might even make federated learning a necessity [Pentland et al., 2021]. Collaboration among clients is most attractive when clients have very different subsets of the combined dataset (data heterogeneity).

Large language models have been shown to achieve remarkable performance across a variety of natural language tasks using few-shot learning, which drastically reduces the number of task-specific training examples needed to adapt the model to a particular application. To further our understanding of the impact of scale on few-shot learning, we trained a 540-billion parameter, densely activated, Transformer language model, which we call Pathways Language Model PaLM. We trained PaLM on 6144 TPU v4 chips using Pathways, a new ML system which enables highly efficient training across multiple TPU Pods. We demonstrate continued benefits of scaling by achieving state-of-the-art few-shot learning results on hundreds of language understanding and generation benchmarks. On a number of these tasks, PaLM 540B achieves breakthrough performance, outperforming the finetuned state-of-the-art on a suite of multi-step reasoning tasks, and outperforming average human performance on the recently released BIG-bench benchmark. A significant number of BIG-bench tasks showed discontinuous improvements from model scale, meaning that performance steeply increased as we scaled to our largest model. PaLM also has strong capabilities in multilingual tasks and source code generation, which we demonstrate on a wide array of benchmarks. We additionally provide a comprehensive analysis on bias and toxicity, and study the extent of training data memorization with respect to model scale. Finally, we discuss the ethical considerations related to large language models and discuss potential mitigation strategies.

Legal literature on machine learning (ML) tends to focus on harms, and thus tends to reason about individual model outcomes and summary error rates. This focus has masked important aspects of ML that are rooted in its reliance on randomness -- namely, stochasticity and non-determinism. While some recent work has begun to reason about the relationship between stochasticity and arbitrariness in legal contexts, the role of non-determinism more broadly remains unexamined. In this paper, we clarify the overlap and differences between these two concepts, and show that the effects of non-determinism, and consequently its implications for the law, become clearer from the perspective of reasoning about ML outputs as distributions over possible outcomes. This distributional viewpoint accounts for randomness by emphasizing the possible outcomes of ML. Importantly, this type of reasoning is not exclusive with current legal reasoning; it complements (and in fact can strengthen) analyses concerning individual, concrete outcomes for specific automated decisions. By illuminating the important role of non-determinism, we demonstrate that ML code falls outside of the cyberlaw frame of treating ``code as law,'' as this frame assumes that code is deterministic. We conclude with a brief discussion of what work ML can do to constrain the potentially harm-inducing effects of non-determinism, and we indicate where the law must do work to bridge the gap between its current individual-outcome focus and the distributional approach that we recommend.

A red-headed woman stands on the moon, her face obscured. Her naked body looks like it belongs on a poster you'd find on a hormonal teenager's bedroom wall -- that is, until you reach her torso, where three arms spit out of her shoulders. AI-powered systems like Stable Diffusion, which translate text prompts into pictures, have been used by brands and artists to create concept images, award-winning (albeit controversial) prints and full-blown marketing campaigns. But some users, intent on exploring the systems' murkier side, have been testing them for a different sort of use case: porn. AI porn is about as unsettling and imperfect as you'd expect (that red-head on the moon was likely not generated by someone with an extra arm fetish).

Artificial intelligence (AI) applications, particularly those focused on biometric data gathering, have recently come under another round of scrutiny both in Europe and Canada. The European Commission proposed the AI Liability Directive last week, a set of rules designed to aid redress for people whose privacy was harmed by AI-powered and digital devices like self-driving cars, voice assistants and drones. According to BBC reporting, the Directive may operate alongside the EU's proposed AI Act if successfully turned into law, introducing a "presumption of causality" for those claiming injuries by AI-enabled products. In other words, individuals harmed by these systems would not have to provide technical explanations for how AI systems work but merely show how they have harmed them in practical terms. "The objective of this proposal is to promote the rollout of trustworthy AI to harvest its full benefits for the internal market. It does so by ensuring victims of damage caused by AI obtain equivalent protection to victims of damage caused by products in general," reads the text of the Directive.

Labeling articulated objects in unconstrained settings have a wide variety of applications including entertainment, neuroscience, psychology, ethology, and many fields of medicine. Large offline labeled datasets do not exist for all but the most common articulated object categories (e.g., humans). Hand labeling these landmarks within a video sequence is a laborious task. Learned landmark detectors can help, but can be error-prone when trained from only a few examples. Multi-camera systems that train fine-grained detectors have shown significant promise in detecting such errors, allowing for self-supervised solutions that only need a small percentage of the video sequence to be hand-labeled. The approach, however, is based on calibrated cameras and rigid geometry, making it expensive, difficult to manage, and impractical in real-world scenarios. In this paper, we address these bottlenecks by combining a non-rigid 3D neural prior with deep flow to obtain high-fidelity landmark estimates from videos with only two or three uncalibrated, handheld cameras. With just a few annotations (representing 1-2% of the frames), we are able to produce 2D results comparable to state-of-the-art fully supervised methods, along with 3D reconstructions that are impossible with other existing approaches. Our Multi-view Bootstrapping in the Wild (MBW) approach demonstrates impressive results on standard human datasets, as well as tigers, cheetahs, fish, colobus monkeys, chimpanzees, and flamingos from videos captured casually in a zoo. We release the codebase for MBW as well as this challenging zoo dataset consisting image frames of tail-end distribution categories with their corresponding 2D, 3D labels generated from minimal human intervention.

Due to legal requirements like the European General Data Protection Regulation (GDPR), the California Consumer Privacy Act, and many others, users gain more control over their personal data collected daily. The right to be forgotten is of particular importance, which states that collected data must be deleted when requested. Deleting data is often insufficient to provide real data privacy. This is especially the case if the data was used to train machine learning models since they might expose information about their training data via white-box or even black-box access. Motivated by this, the field of Machine Unlearning and Forgetting gained more and more attention.

In recent years, Explainable AI (xAI) attracted a lot of attention as various countries turned explanations into a legal right. xAI allows for improving models beyond the accuracy metric by, e.g., debugging the learned pattern and demystifying the AI's behavior. The widespread use of xAI brought new challenges. On the one hand, the number of published xAI algorithms underwent a boom, and it became difficult for practitioners to select the right tool. On the other hand, some experiments did highlight how easy data scientists could misuse xAI algorithms and misinterpret their results. To tackle the issue of comparing and correctly using feature importance xAI algorithms, we propose Compare-xAI, a benchmark that unifies all exclusive functional testing methods applied to xAI algorithms. We propose a selection protocol to shortlist non-redundant functional tests from the literature, i.e., each targeting a specific end-user requirement in explaining a model. The benchmark encapsulates the complexity of evaluating xAI methods into a hierarchical scoring of three levels, namely, targeting three end-user groups: researchers, practitioners, and laymen in xAI. The most detailed level provides one score per test. The second level regroups tests into five categories (fidelity, fragility, stability, simplicity, and stress tests). The last level is the aggregated comprehensibility score, which encapsulates the ease of correctly interpreting the algorithm's output in one easy to compare value. Compare-xAI's interactive user interface helps mitigate errors in interpreting xAI results by quickly listing the recommended xAI solutions for each ML task and their current limitations. The benchmark is made available at https://karim-53.github.io/cxai/

First, Elon Musk has been touting that the next big breakthrough for Tesla will entail the development and fielding of a walking robot that resembles humanoid characteristics. At the Tesla AI Day 2021 last year, there was a rather embarrassing "demonstration" of the envisioned robot that involved a person wearing a robotic-looking costume that leaped and danced around on the stage. I say embarrassing because it was one of the most cringe-worthy moments of any AI showcase.