Large Language Models (LLMs) have achieved remarkable advancements, but their monolithic nature presents challenges in terms of scalability, cost, and customization. This paper introduces the Composition of Experts (CoE), a modular compound AI system leveraging multiple expert LLMs. CoE leverages a router to dynamically select the most appropriate expert for a given input, enabling efficient utilization of resources and improved performance. We formulate the general problem of training a CoE and discuss inherent complexities associated with it. We propose a two-step routing approach to address these complexities that first uses a router to classify the input into distinct categories followed by a category-to-expert mapping to obtain desired experts. CoE offers a flexible and cost-effective solution to build compound AI systems. Our empirical evaluation demonstrates the effectiveness of CoE in achieving superior performance with reduced computational overhead. Given that CoE comprises of many expert LLMs it has unique system requirements for cost-effective serving. We present an efficient implementation of CoE leveraging SambaNova SN40L RDUs unique three-tiered memory architecture. CoEs obtained using open weight LLMs Qwen/Qwen2-7B-Instruct, google/gemma-2-9b-it, google/gemma-2-27b-it, meta-llama/Llama-3.1-70B-Instruct and Qwen/Qwen2-72B-Instruct achieve a score of $59.4$ with merely $31$ billion average active parameters on Arena-Hard and a score of $9.06$ with $54$ billion average active parameters on MT-Bench.

Defending large language models against jailbreaks so that they never engage in a broadly-defined set of forbidden behaviors is an open problem. In this paper, we investigate the difficulty of jailbreak-defense when we only want to forbid a narrowly-defined set of behaviors. As a case study, we focus on preventing an LLM from helping a user make a bomb. We find that popular defenses such as safety training, adversarial training, and input/output classifiers are unable to fully solve this problem. In pursuit of a better solution, we develop a transcript-classifier defense which outperforms the baseline defenses we test. However, our classifier defense still fails in some circumstances, which highlights the difficulty of jailbreak-defense even in a narrow domain.

In response to the practical demands of the ``right to be forgotten" and the removal of undesired data, machine unlearning emerges as an essential technique to remove the learned knowledge of a fraction of data points from trained models. However, existing methods suffer from limitations such as insufficient methodological support, high computational complexity, and significant memory demands. In this work, we propose the concepts of knowledge vaporization and concentration to selectively erase learned knowledge from specific data points while maintaining representations for the remaining data. Utilizing the Siamese networks, we exemplify the proposed concepts and develop an efficient method for machine unlearning. Our proposed Siamese unlearning method does not require additional memory overhead and full access to the remaining dataset. Extensive experiments conducted across multiple unlearning scenarios showcase the superiority of Siamese unlearning over baseline methods, illustrating its ability to effectively remove knowledge from forgetting data, enhance model utility on remaining data, and reduce susceptibility to membership inference attacks.

Recent privacy regulations (e.g., GDPR) grant data subjects the `Right to Be Forgotten' (RTBF) and mandate companies to fulfill data erasure requests from data subjects. However, companies encounter great challenges in complying with the RTBF regulations, particularly when asked to erase specific training data from their well-trained predictive models. While researchers have introduced machine unlearning methods aimed at fast data erasure, these approaches often overlook maintaining model performance (e.g., accuracy), which can lead to financial losses and non-compliance with RTBF obligations. This work develops a holistic machine learning-to-unlearning framework, called Ensemble-based iTerative Information Distillation (ETID), to achieve efficient data erasure while preserving the business value of predictive models. ETID incorporates a new ensemble learning method to build an accurate predictive model that can facilitate handling data erasure requests. ETID also introduces an innovative distillation-based unlearning method tailored to the constructed ensemble model to enable efficient and effective data erasure. Extensive experiments demonstrate that ETID outperforms various state-of-the-art methods and can deliver high-quality unlearned models with efficiency. We also highlight ETID's potential as a crucial tool for fostering a legitimate and thriving market for data and predictive services.

This study examines the capabilities of advanced Large Language Models (LLMs), particularly the o1 model, in the context of literary analysis. The outputs of these models are compared directly to those produced by graduate-level human participants. By focusing on two Nobel Prize-winning short stories, 'Nine Chapters' by Han Kang, the 2024 laureate, and 'Friendship' by Jon Fosse, the 2023 laureate, the research explores the extent to which AI can engage with complex literary elements such as thematic analysis, intertextuality, cultural and historical contexts, linguistic and structural innovations, and character development. Given the Nobel Prize's prestige and its emphasis on cultural, historical, and linguistic richness, applying LLMs to these works provides a deeper understanding of both human and AI approaches to interpretation. The study uses qualitative and quantitative evaluations of coherence, creativity, and fidelity to the text, revealing the strengths and limitations of AI in tasks typically reserved for human expertise. While LLMs demonstrate strong analytical capabilities, particularly in structured tasks, they often fall short in emotional nuance and coherence, areas where human interpretation excels. This research underscores the potential for human-AI collaboration in the humanities, opening new opportunities in literary studies and beyond.

The REAIM 2024 Blueprint for Action states that AI applications in the military domain should be ethical and human-centric and that humans must remain responsible and accountable for their use and effects. Developing rigorous test and evaluation, verification and validation (TEVV) frameworks will contribute to robust oversight mechanisms. TEVV in the development and deployment of AI systems needs to involve human users throughout the lifecycle. Traditional human-centred test and evaluation methods from human factors need to be adapted for deployed AI systems that require ongoing monitoring and evaluation. The language around AI-enabled systems should be shifted to inclusion of the human(s) as a component of the system. Standards and requirements supporting this adjusted definition are needed, as are metrics and means to evaluate them. The need for dialogue between technologists and policymakers on human-centred TEVV will be evergreen, but dialogue needs to be initiated with an objective in mind for it to be productive. Development of TEVV throughout system lifecycle is critical to support this evolution including the issue of human scalability and impact on scale of achievable testing. Communication between technical and non technical communities must be improved to ensure operators and policy-makers understand risk assumed by system use and to better inform research and development. Test and evaluation in support of responsible AI deployment must include the effect of the human to reflect operationally realised system performance. Means of communicating the results of TEVV to those using and making decisions regarding the use of AI based systems will be key in informing risk based decisions regarding use.

Elon Musk's attorneys filed for an injunction against OpenAI and Microsoft on Friday accusing the two of anticompetitive practices and seeking to stop OpenAI's conversion to a for-profit company. The filing, spotted by TechCrunch, also names OpenAI CEO Sam Altman, OpenAI President Greg Brockman, Microsoft's Dee Templeton and LinkedIn co-founder Reid Hoffman as defendants. Musk first sued OpenAI earlier this year for allegedly violating its founding mission of building AI "for the benefit of humanity," but withdrew the lawsuit a few months later. He then filed another lawsuit against OpenAI in a California federal court in August, and recently added Microsoft as a defendant. The new motion accuses OpenAI and Microsoft of telling investors not to fund OpenAI's competitors, such as Musk's xAI, of "benefitting from wrongfully obtained competitively sensitive information or coordination" through its relationship with Microsoft, and other alleged antitrust violations.

Large language models (LLMs) integrated into multistep agent systems enable complex decision-making processes across various applications. However, their outputs often lack reliability, making uncertainty estimation crucial. Existing uncertainty estimation methods primarily focus on final-step outputs, which fail to account for cumulative uncertainty over the multistep decision-making process and the dynamic interactions between agents and their environments. To address these limitations, we propose SAUP (Situation Awareness Uncertainty Propagation), a novel framework that propagates uncertainty through each step of an LLM-based agent's reasoning process. SAUP incorporates situational awareness by assigning situational weights to each step's uncertainty during the propagation. Our method, compatible with various one-step uncertainty estimation techniques, provides a comprehensive and accurate uncertainty measure. Extensive experiments on benchmark datasets demonstrate that SAUP significantly outperforms existing state-of-the-art methods, achieving up to 20% improvement in AUROC.

In this paper, we introduce QABISAR, a novel framework for statutory article retrieval, to overcome the semantic mismatch problem when modeling each query-article pair in isolation, making it hard to learn representation that can effectively capture multi-faceted information. QABISAR leverages bipartite interactions between queries and articles to capture diverse aspects inherent in them. Further, we employ knowledge distillation to transfer enriched query representations from the graph network into the query bi-encoder, to capture the rich semantics present in the graph representations, despite absence of graph-based supervision for unseen queries during inference. Our experiments on a real-world expert-annotated dataset demonstrate its effectiveness.

Large language models (LLMs) have become increasingly pivotal in various domains due the recent advancements in their performance capabilities. However, concerns persist regarding biases in LLMs, including gender, racial, and cultural biases derived from their training data. These biases raise critical questions about the ethical deployment and societal impact of LLMs. Acknowledging these concerns, this study investigates whether LLMs accurately reflect cross-cultural variations and similarities in moral perspectives. In assessing whether the chosen LLMs capture patterns of divergence and agreement on moral topics across cultures, three main methods are employed: (1) comparison of model-generated and survey-based moral score variances, (2) cluster alignment analysis to evaluate the correspondence between country clusters derived from model-generated moral scores and those derived from survey data, and (3) probing LLMs with direct comparative prompts. All three methods involve the use of systematic prompts and token pairs designed to assess how well LLMs understand and reflect cultural variations in moral attitudes. The findings of this study indicate overall variable and low performance in reflecting cross-cultural differences and similarities in moral values across the models tested, highlighting the necessity for improving models' accuracy in capturing these nuances effectively. The insights gained from this study aim to inform discussions on the ethical development and deployment of LLMs in global contexts, emphasizing the importance of mitigating biases and promoting fair representation across diverse cultural perspectives.