The lawsuit goes to the heart of a major unresolved question hanging over the surge in "generative" AI tools such as chatbots and image generators. The technology works by ingesting billions of words from the open internet and learning to build inferences between them. After consuming enough data, the resulting "large language models" can predict what to say in response to a prompt, giving them the ability to write poetry, have complex conversations and pass professional exams. But the humans who wrote those billions of words never signed off on having a company such as OpenAI use them for its own profit.

Paedophiles are using artificial intelligence (AI) to create life-like child sexual abuse images, it has been claimed. A shocking investigation has alleged that depictions of babies and toddlers being raped are among numerous images produced by abusers using Stable Diffusion. While the image generation software is intended to make artwork, the AI's use of word prompts allows for any desired image to be formed. The BBC claimed these are then being sold and traded on the Japanese site of Pixiv, with accounts often leading to more explicit content on the US-based Patreon. 'Since AI-generated images became possible, there has been this huge flood… it's not just very young girls, they're [paedophiles] talking about toddlers,' Journalist, Olivia Sheepshanks, told the publication. The new report come just a month after the AI platform Midjourney was also found to transform real photos of children into sexualised images.

Among democratic nations, Britain is at the forefront of using live facial recognition, with courts and regulators signing off on its use. The police in London and Cardiff are experimenting with the technology to identify wanted criminals as they walk down the street. In May, it was used to scan the crowds at the coronation of King Charles III. But the use by retailers has drawn criticism as a disproportionate solution for minor crimes. Individuals have little way of knowing they are on the watchlist or how to appeal.

As machine learning (ML) based systems are adopted in domains such as law enforcement, criminal justice, finance, hiring and admissions, ensuring the fairness of ML aided decision-making is becoming increasingly important. In this paper, we focus on the problem of fair classification, and introduce a novel min-max F-divergence regularization framework for learning fair classification models while preserving high accuracy. Our framework consists of two trainable networks, namely, a classifier network and a bias/fairness estimator network, where the fairness is measured using the statistical notion of F-divergence. We show that F-divergence measures possess convexity and differentiability properties, and their variational representation make them widely applicable in practical gradient based training methods. The proposed framework can be readily adapted to multiple sensitive attributes and for high dimensional datasets. We study the F-divergence based training paradigm for two types of group fairness constraints, namely, demographic parity and equalized odds. We present a comprehensive set of experiments for several real-world data sets arising in multiple domains (including COMPAS, Law Admissions, Adult Income, and CelebA datasets). To quantify the fairness-accuracy tradeoff, we introduce the notion of fairness-accuracy receiver operating characteristic (FA-ROC) and a corresponding \textit{low-bias} FA-ROC, which we argue is an appropriate measure to evaluate different classifiers. In comparison to several existing approaches for learning fair classifiers (including pre-processing, post-processing and other regularization methods), we show that the proposed F-divergence based framework achieves state-of-the-art performance with respect to the trade-off between accuracy and fairness.

We study causal inference and efficient estimation for the expected number of recurrent events in the presence of a terminal event. We define our estimand as the vector comprising both the expected number of recurrent events and the failure survival function evaluated along a sequence of landmark times. We identify the estimand in the presence of right-censoring and causal selection as an observed data functional under coarsening at random, derive the nonparametric efficiency bound, and propose a multiply-robust estimator that achieves the bound and permits nonparametric estimation of nuisance parameters. Throughout, no absolute continuity assumption is made on the underlying probability distributions of failure, censoring, or the observed data. Additionally, we derive the class of influence functions when the coarsening distribution is known and review how published estimators may belong to the class. Along the way, we highlight some interesting inconsistencies in the causal lifetime analysis literature.

Automatically generating textual content with desired attributes is an ambitious task that people have pursued long. Existing works have made a series of progress in incorporating unimodal controls into language models (LMs), whereas how to generate controllable sentences with multimodal signals and high efficiency remains an open question. To tackle the puzzle, we propose a new paradigm of zero-shot controllable text generation with multimodal signals (\textsc{ZeroGen}). Specifically, \textsc{ZeroGen} leverages controls of text and image successively from token-level to sentence-level and maps them into a unified probability space at decoding, which customizes the LM outputs by weighted addition without extra training. To achieve better inter-modal trade-offs, we further introduce an effective dynamic weighting mechanism to regulate all control weights. Moreover, we conduct substantial experiments to probe the relationship of being in-depth or in-width between signals from distinct modalities. Encouraging empirical results on three downstream tasks show that \textsc{ZeroGen} not only outperforms its counterparts on captioning tasks by a large margin but also shows great potential in multimodal news generation with a higher degree of control. Our code will be released at https://github.com/ImKeTT/ZeroGen.

Negation is a fundamental aspect of natural language, playing a critical role in communication and comprehension. Our study assesses the negation detection performance of Generative Pre-trained Transformer (GPT) models, specifically GPT-2, GPT-3, GPT-3.5, and GPT-4. We focus on the identification of negation in natural language using a zero-shot prediction approach applied to our custom xNot360 dataset. Our approach examines sentence pairs labeled to indicate whether the second sentence negates the first. Our findings expose a considerable performance disparity among the GPT models, with GPT-4 surpassing its counterparts and GPT-3.5 displaying a marked performance reduction. The overall proficiency of the GPT models in negation detection remains relatively modest, indicating that this task pushes the boundaries of their natural language understanding capabilities. We not only highlight the constraints of GPT models in handling negation but also emphasize the importance of logical reliability in high-stakes domains such as healthcare, science, and law.

Large Language Models (LLMs) have shown the potential to revolutionize natural language processing tasks in various domains, sparking great interest in vertical-specific large models. However, unlike proprietary models such as BloombergGPT and FinGPT, which have leveraged their unique data accumulations to make strides in the finance domain, there hasn't not many similar large language models in the Chinese legal domain to facilitate its digital transformation. In this paper, we propose an open-source legal large language model named ChatLaw. Due to the importance of data quality, we carefully designed a legal domain fine-tuning dataset. Additionally, to overcome the problem of model hallucinations in legal data screening during reference data retrieval, we introduce a method that combines vector database retrieval with keyword retrieval to effectively reduce the inaccuracy of relying solely on vector database retrieval. Furthermore, we propose a self-attention method to enhance the ability of large models to overcome errors present in reference data, further optimizing the issue of model hallucinations at the model level and improving the problem-solving capabilities of large models. We also open-sourced our model and part of the data at https://github.com/PKU-YuanGroup/ChatLaw.

The passage retrieval problem, which is of central The principal limitation to this approach is its dependence importance in search engine optimization and text on explicit term matches between the analytics, entails the following: given a set of documents query and the context. In many cases, the correct and a query, determine which document best context-query pair may have no words in common.

We study the problem of learning mixtures of Gaussians with censored data. Statistical learning with censored data is a classical problem, with numerous practical applications, however, finite-sample guarantees for even simple latent variable models such as Gaussian mixtures are missing. Formally, we are given censored data from a mixture of univariate Gaussians $$ \sum_{i=1}^k w_i \mathcal{N}(\mu_i,\sigma^2), $$ i.e. the sample is observed only if it lies inside a set $S$. The goal is to learn the weights $w_i$ and the means $\mu_i$. We propose an algorithm that takes only $\frac{1}{\varepsilon^{O(k)}}$ samples to estimate the weights $w_i$ and the means $\mu_i$ within $\varepsilon$ error.