Culture fundamentally shapes people's reasoning, behavior, and communication. Generative artificial intelligence (AI) technologies may cause a shift towards a dominant culture. As people increasingly use AI to expedite and even automate various professional and personal tasks, cultural values embedded in AI models may bias authentic expression. We audit large language models for cultural bias, comparing their responses to nationally representative survey data, and evaluate country-specific prompting as a mitigation strategy. We find that GPT-4, 3.5 and 3 exhibit cultural values resembling English-speaking and Protestant European countries. Our mitigation strategy reduces cultural bias in recent models but not for all countries/territories. To avoid cultural bias in generative AI, especially in high-stakes contexts, we suggest using culture matching and ongoing cultural audits.

The lack of explainability using relevant clinical knowledge hinders the adoption of Artificial Intelligence-powered analysis of unstructured clinical dialogue. A wealth of relevant, untapped Mental Health (MH) data is available in online communities, providing the opportunity to address the explainability problem with substantial potential impact as a screening tool for both online and offline applications. We develop a method to enhance attention in popular transformer models and generate clinician-understandable explanations for classification by incorporating external clinical knowledge. Inspired by how clinicians rely on their expertise when interacting with patients, we leverage relevant clinical knowledge to model patient inputs, providing meaningful explanations for classification. This will save manual review time and engender trust. We develop such a system in the context of MH using clinical practice guidelines (CPG) for diagnosing depression, a mental health disorder of global concern. We propose an application-specific language model called ProcesS knowledge-infused cross ATtention (PSAT), which incorporates CPGs when computing attention. Through rigorous evaluation on three expert-curated datasets related to depression, we demonstrate application-relevant explainability of PSAT. PSAT also surpasses the performance of nine baseline models and can provide explanations where other baselines fall short. We transform a CPG resource focused on depression, such as the Patient Health Questionnaire (e.g. PHQ-9) and related questions, into a machine-readable ontology using SNOMED-CT. With this resource, PSAT enhances the ability of models like GPT-3.5 to generate application-relevant explanations.

Recent advancements in language models have showcased human-comparable performance in academic entrance exams. However, existing studies often overlook questions that require the integration of visual comprehension, thus compromising the full spectrum and complexity inherent in real-world scenarios. To address this gap, we present a comprehensive framework to evaluate language models on entrance exams, which incorporates both textual and visual elements. We evaluate the two most recent editions of Exame Nacional do Ensino M\'edio (ENEM), the main standardized entrance examination adopted by Brazilian universities. Our study not only reaffirms the capabilities of GPT-4 as the state of the art for handling complex multidisciplinary questions, but also pioneers in offering a realistic assessment of multimodal language models on Portuguese examinations. One of the highlights is that text captions transcribing visual content outperform the direct use of images, suggesting that the vision model has room for improvement. Yet, despite improvements afforded by images or captions, mathematical questions remain a challenge for these state-of-the-art models. The code and data used on experiments are available at https://github.com/piresramon/gpt-4-enem.

Large Language Models (LLM) have made significant advances in the recent past becoming more mainstream in Artificial Intelligence (AI) enabled human-facing applications. However, LLMs often generate stereotypical output inherited from historical data, amplifying societal biases and raising ethical concerns. This work introduces i) the Multi-Grain Stereotype Dataset, which includes 52,751 instances of gender, race, profession and religion stereotypic text and ii) a novel stereotype classifier for English text. We design several experiments to rigorously test the proposed model trained on the novel dataset. Our experiments show that training the model in a multi-class setting can outperform the one-vs-all binary counterpart. Consistent feature importance signals from different eXplainable AI tools demonstrate that the new model exploits relevant text features. We utilise the newly created model to assess the stereotypic behaviour of the popular GPT family of models and observe the reduction of bias over time. In summary, our work establishes a robust and practical framework for auditing and evaluating the stereotypic bias in LLM.

Multimodal reasoning is a challenging task that requires models to reason across multiple modalities to answer questions. Existing approaches have made progress by incorporating language and visual modalities into a two-stage reasoning framework, separating rationale generation from answer inference. However, these approaches often fall short due to the inadequate quality of the generated rationales. In this work, we delve into the importance of rationales in model reasoning. We observe that when rationales are completely accurate, the model's accuracy significantly improves, highlighting the need for high-quality rationale generation. Motivated by this, we propose MC-CoT, a self-consistency training strategy that generates multiple rationales and answers, subsequently selecting the most accurate through a voting process. This approach not only enhances the quality of generated rationales but also leads to more accurate and robust answers. Through extensive experiments, we demonstrate that our approach significantly improves model performance across various benchmarks. Remarkably, we show that even smaller base models, when equipped with our proposed approach, can achieve results comparable to those of larger models, illustrating the potential of our approach in harnessing the power of rationales for improved multimodal reasoning. The code is available at https://github.com/chengtan9907/mc-cot.

Digital art portfolios serve as impactful mediums for artists to convey their visions, weaving together visuals, audio, interactions, and narratives. However, without technical backgrounds, design students often find it challenging to translate creative ideas into tangible codes and designs, given the lack of tailored resources for the non-technical, academic support in art schools, and a comprehensive guiding tool throughout the mentally demanding process. Recognizing the role of companionship in code learning and leveraging generative AI models' capabilities in supporting creative tasks, we present PortfolioMentor, a coding companion chatbot for IDEs. This tool guides and collaborates with students through proactive suggestions and responsible Q&As for learning, inspiration, and support. In detail, the system starts with the understanding of the task and artist's visions, follows the co-creation of visual illustrations, audio or music suggestions and files, click-scroll effects for interactions, and creative vision conceptualization, and finally synthesizes these facets into a polished interactive digital portfolio.

This paper bridges the gap between mathematical heuristic strategies learned from Deep Reinforcement Learning (DRL) in automated agent negotiation, and comprehensible, natural language explanations. Our aim is to make these strategies more accessible to non-experts. By leveraging traditional Natural Language Processing (NLP) techniques and Large Language Models (LLMs) equipped with Transformers, we outline how parts of DRL strategies composed of parts within strategy templates can be transformed into user-friendly, human-like English narratives. To achieve this, we present a top-level algorithm that involves parsing mathematical expressions of strategy templates, semantically interpreting variables and structures, generating rule-based primary explanations, and utilizing a Generative Pre-trained Transformer (GPT) model to refine and contextualize these explanations. Subsequent customization for varied audiences and meticulous validation processes in an example illustrate the applicability and potential of this approach.

End users face a choice between privacy and efficiency in current Large Language Model (LLM) service paradigms. In cloud-based paradigms, users are forced to compromise data locality for generation quality and processing speed. Conversely, edge device paradigms maintain data locality but fail to deliver satisfactory performance. In this work, we propose a novel LLM service paradigm that distributes privacy-sensitive computation on edge devices and shared computation in the cloud. Only activations are transmitted between the central cloud and edge devices to ensure data locality. Our core innovation, PrivateLoRA, addresses the challenging communication overhead by exploiting the low rank of residual activations, achieving over 95% communication reduction. Consequently, PrivateLoRA effectively maintains data locality and is extremely resource efficient. Under standard 5G networks, PrivateLoRA achieves throughput over 300% of device-only solutions for 7B models and over 80% of an A100 GPU for 33B models. PrivateLoRA also provides tuning performance comparable to LoRA for advanced personalization. Our approach democratizes access to state-of-the-art generative AI for edge devices, paving the way for more tailored LLM experiences for the general public. To our knowledge, our proposed framework is the first efficient and privacy-preserving LLM solution in the literature.

Large language models (LLMs) are capable of answering knowledge-intensive complex questions with chain-of-thought (CoT) reasoning. However, they tend to generate factually incorrect reasoning steps when the required knowledge is not available or up-to-date in models' parameters. Recent works turn to retrieving external knowledge to augment CoT reasoning. Despite being promising, these chain-based methods suffer from: 1) Negative retrieval. Unnecessary or incorrect retrieval may mislead the reasoning; 2) Limited sight. Lacking the ability to look backward or forward, a local error in one step will propagate along the chain. In this paper, we propose a novel approach: Probabilistic Tree-of-thought Reasoning (ProbTree). First, LLMs translate a complex question into a query tree, in which each non-root node denotes a sub-question of its parent node. Then, probabilistic reasoning is conducted over the tree, by solving questions from leaf to root considering the confidence of both question decomposing and answering. During reasoning, for leaf nodes, LLMs choose a more confident answer from Closed-book QA that employs parametric knowledge and Open-book QA that employs retrieved external knowledge, thus eliminating the negative retrieval problem. For non-leaf nodes, with the hierarchical structure, LLMs have broader sights and are able to globally reason with the information from child nodes, thus recovering from local errors. The experiments on three Complex QA datasets under the open-domain setting show that our approach outperforms SOTA methods significantly, demonstrating the effect of probabilistic tree-of-thought reasoning.

In the pursuit of Artificial General Intelligence (AGI), the integration of vision in language models has marked a significant milestone. The advent of vision-language models (MLLMs) like GPT-4V have expanded AI applications, aligning with the multi-modal capabilities of the human brain. However, evaluating the efficacy of MLLMs poses a substantial challenge due to the subjective nature of tasks that lack definitive answers. Existing automatic evaluation methodologies on multi-modal large language models rely on objective queries that have standard answers, inadequately addressing the nuances of creative and associative multi-modal tasks. To address this, we introduce MLLM-Bench, an innovative benchmark inspired by Vicuna, spanning a diverse array of scenarios, including Perception, Understanding, Applying, Analyzing, Evaluating, and Creation along with the ethical consideration. MLLM-Bench is designed to reflect user experience more accurately and provide a more holistic assessment of model performance. Comparative evaluations indicate a significant performance gap between existing open-source models and GPT-4V. We posit that MLLM-Bench will catalyze progress in the open-source community towards developing user-centric vision-language models that meet a broad spectrum of real-world applications. See online leaderboard in \url{https://mllm-bench.llmzoo.com}.