The accompanying shadow represents the standard deviation ( Std). Recent research has extended beyond assessing the performance of Large Language Models (LLMs) to examining their characteristics from a psychological standpoint, acknowledging the necessity of understanding their behavioral characteristics. The administration of personality tests to LLMs has emerged as a noteworthy area in this context. However, the suitability of employing psychological scales, initially devised for humans, on LLMs is a matter of ongoing debate. Our study aims to determine the reliability of applying personality assessments to LLMs, explicitly investigating whether LLMs demonstrate consistent personality traits. Analyzing responses under 2,500 settings reveals that gpt-3.5-turbo Furthermore, our research explores the potential of gpt-3.5-turbo to emulate diverse personalities and represent various groups--a capability increasingly sought after in social sciences for substituting human participants with LLMs to reduce costs. Our findings reveal that LLMs have the potential to represent different personalities with specific prompt instructions. By shedding light on the personalization of LLMs, our study endeavors to pave the way for future explorations in this field. Wenxiang Jiao is the corresponding author. The recent emergence of Large Language Models (LLMs) marks a significant advancement in the field of Artificial Intelligence (AI), representing a notable milestone.

Post-hoc global/local feature attribution methods are progressively being employed to understand the decisions of complex machine learning models. Yet, because of limited amounts of data, it is possible to obtain a diversity of models with good empirical performance but that provide very different explanations for the same prediction, making it hard to derive insight from them. In this work, instead of aiming at reducing the under-specification of model explanations, we fully embrace it and extract logical statements about feature attributions that are consistent across all models with good empirical performance (i.e. all models in the Rashomon Set). We show that partial orders of local/global feature importance arise from this methodology enabling more nuanced interpretations by allowing pairs of features to be incomparable when there is no consensus on their relative importance. We prove that every relation among features present in these partial orders also holds in the rankings provided by existing approaches. Finally, we present three use cases employing hypothesis spaces with tractable Rashomon Sets (Additive models, Kernel Ridge, and Random Forests) and show that partial orders allow one to extract consistent local and global interpretations of models despite their under-specification.

While previous lawsuits claiming intellectual property violations by AI companies have come from artists and writers, the Times is the first American news organization to sue the companies, alleging that OpenAI and Microsoft used millions of their articles to train digital chatbots that now compete with the publication. While the case does not specify the revenue the Times has lost to new robot rivals, the suit argues that the tech companies' unauthorized use of the newspaper's images and written work deprives it of income from "subscriptions, licensing, advertising, and affiliates." The complaint asks that the AI companies be held accountable for "billions of dollars in statutory and actual" damages, citing several examples where the program lifted excerpts from the paper's stories verbatim. "Defendants have refused to recognize this protection." You can read the full legal complaint here.

The New York Times is suing OpenAI and Microsoft for using published news articles to train its artificial intelligence chatbots without an agreement that compensates it for its intellectual property. The NYT did not specify how much it seeks in payout from the companies but that "this action seeks to hold them responsible for the billions of dollars in statutory and actual damages." The NYT claims that OpenAI and Microsoft, the makers of Chat GPT and Copilot, "seek to free-ride on The Times's massive investment in its journalism" without having any licensing agreements. In one part of the complaint, the NYT highlights that its domain (www.nytimes.com) It alleges more than 66 million records, ranging from breaking news articles to op-eds, published across the NYT websites and other affiliated brands were used to train the AI models.

We may start to find out in 2024. In the US, there are nine separate class-action lawsuits against AI firms, including OpenAI, Meta and Google parent company Alphabet. One of those, Getty Images vs Stability AI, is also being tried in the High Court in London.

The Times said OpenAI and Microsoft are advancing their technology through the "unlawful use of The Times's work to create artificial intelligence products that compete with it" and "threatens The Times's ability to provide that service". Through their AI chatbots, the companies "seek to free-ride on The Times's massive investment in its journalism by using it to build substitutive products without permission or payment", the lawsuit said. The Times, one of the most respected news organisations in the United States, is seeking damages as well as an order that the companies stop using its content – and destroy data already harvested. While no sum is specifically requested, the Times alleged that the infringement could have cost "billions of dollars in statutory and actual damages". With the suit, The New York Times chose a more confrontational approach to the sudden rise of AI chatbots, in contrast to other media groups, such as Germany's Axel Springer or The Associated Press, which have struck content deals with OpenAI. Microsoft, the world's second biggest company by market capitalisation, is a major investor in OpenAI and swiftly implemented the powers of AI in its own products after the release of ChatGPT last year.

In a complaint filed Wednesday, the Times said the technology companies exploited its content without permission to create their AI products, including OpenAI's humanlike chatbot ChatGPT and Microsoft's Copilot. The tools were trained on millions of pieces of Times content, the suit said, and draw on that material to serve up answers to users' prompts.

Censoring is the central problem in survival analysis where either the time-to-event (for instance, death), or the time-tocensoring (such as loss of follow-up) is observed for each sample. The majority of existing machine learning-based survival analysis methods assume that survival is conditionally independent of censoring given a set of covariates; an assumption that cannot be verified since only marginal distributions is available from the data. The existence of dependent censoring, along with the inherent bias in current estimators has been demonstrated in a variety of applications, accentuating the need for a more nuanced approach. However, existing methods that adjust for dependent censoring require practitioners to specify the ground truth copula. This requirement poses a significant challenge for practical applications, as model misspecification can lead to substantial bias. In this work, we propose a flexible deep learning-based survival analysis method that simultaneously accommodate for dependent censoring and eliminates the requirement for specifying the ground truth copula. We theoretically prove the identifiability of our model under a broad family of copulas and survival distributions. Experiments results from a wide range of datasets demonstrate that our approach successfully discerns the underlying dependency structure and significantly reduces survival estimation bias when compared to existing methods.

Abstract--The recent trend of large language models (LLMs) is to increase the scale of both model size (a.k.a the number of parameters) and dataset to achieve better generative ability, which is definitely proved by a lot of work such as the famous GPT and Llama. However, large models often involve massive computational costs, and practical applications cannot afford such high prices. However, the method of constructing a strong model architecture for LLMs is rarely discussed. We first analyze the state-of-the-art language model architectures and observe the feature collapse problem. Based on the theoretical analysis, we propose that the nonlinearity is also very important for language models, which is usually studied in convolutional neural networks for vision tasks. The series informed activation function is then introduced with tiny calculations that can be ignored, and an augmented shortcut is further used to enhance the model nonlinearity. We then demonstrate that the proposed approach is significantly effective for enhancing the model nonlinearity through carefully designed ablations; thus, we present a new efficient model architecture for establishing modern, namely, PanGu- π . Experiments are then conducted using the same dataset and training strategy to compare PanGu- π with state-of-the-art LLMs. The results show that PanGu- π -7B can achieve a comparable performance to that of benchmarks with about 10% inference speed-up, and PanGu- π -1B can achieve state-of-the-art performance in terms of accuracy and efficiency. In addition, we have deployed PanGu- π -7B in the high-value domains of finance and law, developing an LLM named YunShan for practical application. The results show that YunShan can surpass other models with similar scales on benchmarks. As shown in Figure 1, our translation, text summarization, and dialogue.

Current datasets for unwanted social bias auditing are limited to studying protected demographic features such as race and gender. In this work, we introduce a comprehensive benchmark that is meant to capture the amplification of social bias, via stigmas, in generative language models. Taking inspiration from social science research, we start with a documented list of 93 US-centric stigmas and curate a question-answering (QA) dataset which involves simple social situations. Our benchmark, SocialStigmaQA, contains roughly 10K prompts, with a variety of prompt styles, carefully constructed to systematically test for both social bias and model robustness. We present results for SocialStigmaQA with two open source generative language models and we find that the proportion of socially biased output ranges from 45% to 59% across a variety of decoding strategies and prompting styles. We demonstrate that the deliberate design of the templates in our benchmark (e.g., adding biasing text to the prompt or using different verbs that change the answer that indicates bias) impacts the model tendencies to generate socially biased output. Additionally, through manual evaluation, we discover problematic patterns in the generated chain-of-thought output that range from subtle bias to lack of reasoning. Warning: This paper contains examples of text which are toxic, biased, and potentially harmful.