Although supervised machine learning is popular for information extraction from clinical notes, creating large annotated datasets requires extensive domain expertise and is time-consuming. Meanwhile, large language models (LLMs) have demonstrated promising transfer learning capability. In this study, we explored whether recent LLMs can reduce the need for large-scale data annotations. We curated a manually-labeled dataset of 769 breast cancer pathology reports, labeled with 13 categories, to compare zero-shot classification capability of the GPT-4 model and the GPT-3.5 model with supervised classification performance of three model architectures: random forests classifier, long short-term memory networks with attention (LSTM-Att), and the UCSF-BERT model. Across all 13 tasks, the GPT-4 model performed either significantly better than or as well as the best supervised model, the LSTM-Att model (average macro F1 score of 0.83 vs. 0.75). On tasks with high imbalance between labels, the differences were more prominent. Frequent sources of GPT-4 errors included inferences from multiple samples and complex task design. On complex tasks where large annotated datasets cannot be easily collected, LLMs can reduce the burden of large-scale data labeling. However, if the use of LLMs is prohibitive, the use of simpler supervised models with large annotated datasets can provide comparable results. LLMs demonstrated the potential to speed up the execution of clinical NLP studies by reducing the need for curating large annotated datasets. This may result in an increase in the utilization of NLP-based variables and outcomes in observational clinical studies.

Large language models like GPT-3.5-turbo and GPT-4 hold promise for healthcare professionals, but they may inadvertently inherit biases during their training, potentially affecting their utility in medical applications. Despite few attempts in the past, the precise impact and extent of these biases remain uncertain. Through both qualitative and quantitative analyses, we find that these models tend to project higher costs and longer hospitalizations for White populations and exhibit optimistic views in challenging medical scenarios with much higher survival rates. These biases, which mirror real-world healthcare disparities, are evident in the generation of patient backgrounds, the association of specific diseases with certain races, and disparities in treatment recommendations, etc. Our findings underscore the critical need for future research to address and mitigate biases in language models, especially in critical healthcare applications, to ensure fair and accurate outcomes for all patients.

Large Language Models (LLMs) have recently showcased remarkable reasoning abilities. However, larger models often surpass their smaller counterparts in reasoning tasks, posing the challenge of effectively transferring these capabilities from larger models. Existing approaches heavily rely on extensive fine-tuning data or continuous interactions with a superior teacher LLM during inference. We introduce a principle-based teacher-student framework called ``Teaching via Principle Discovery'' (TPD) to address these limitations. Inspired by human learning mechanisms, TPD mimics the interaction between a teacher and a student using a principle-based approach. The teacher LLM generates problem-solving instructions and corrective principles based on the student LLM's errors. These principles guide the refinement of instructions and the selection of instructive examples from a validation set. This enables the student model to learn from both the teacher's guidance and its own mistakes. Once the student model begins making inferences, TPD requires no further intervention from the teacher LLM or humans. Through extensive experiments across eight reasoning tasks, we demonstrate the effectiveness of TPD. Compared to standard chain-of-thought prompting, TPD significantly improves the student model's performance, achieving $6.2\%$ improvement on average.

For large language models (LLMs) to be trusted by humans they need to be well-calibrated in the sense that they can accurately assess and communicate how likely it is that their predictions are correct. Recent work has focused on the quality of internal LLM confidence assessments, but the question remains of how well LLMs can communicate this internal model confidence to human users. This paper explores the disparity between external human confidence in an LLM's responses and the internal confidence of the model. Through experiments involving multiple-choice questions, we systematically examine human users' ability to discern the reliability of LLM outputs. Our study focuses on two key areas: (1) assessing users' perception of true LLM confidence and (2) investigating the impact of tailored explanations on this perception. The research highlights that default explanations from LLMs often lead to user overestimation of both the model's confidence and its' accuracy. By modifying the explanations to more accurately reflect the LLM's internal confidence, we observe a significant shift in user perception, aligning it more closely with the model's actual confidence levels. This adjustment in explanatory approach demonstrates potential for enhancing user trust and accuracy in assessing LLM outputs. The findings underscore the importance of transparent communication of confidence levels in LLMs, particularly in high-stakes applications where understanding the reliability of AI-generated information is essential.

Root Cause Analysis (RCA) plays a pivotal role in the incident diagnosis process for cloud services, requiring on-call engineers to identify the primary issues and implement corrective actions to prevent future recurrences. Improving the incident RCA process is vital for minimizing service downtime, customer impact and manual toil. Recent advances in artificial intelligence have introduced state-of-the-art Large Language Models (LLMs) like GPT-4, which have proven effective in tackling various AIOps problems, ranging from code authoring to incident management. Nonetheless, the GPT-4 model's immense size presents challenges when trying to fine-tune it on user data because of the significant GPU resource demand and the necessity for continuous model fine-tuning with the emergence of new data. To address the high cost of fine-tuning LLM, we propose an in-context learning approach for automated root causing, which eliminates the need for fine-tuning. We conduct extensive study over 100,000 production incidents, comparing several large language models using multiple metrics. The results reveal that our in-context learning approach outperforms the previous fine-tuned large language models such as GPT-3 by an average of 24.8\% across all metrics, with an impressive 49.7\% improvement over the zero-shot model. Moreover, human evaluation involving actual incident owners demonstrates its superiority over the fine-tuned model, achieving a 43.5\% improvement in correctness and an 8.7\% enhancement in readability. The impressive results demonstrate the viability of utilizing a vanilla GPT model for the RCA task, thereby avoiding the high computational and maintenance costs associated with a fine-tuned model.

Emotional user utterances frequently occur during interactions with dialogue systems [19]. In open-domain dialogues, users directly share personal and emotional experiences [10]. In task-oriented dialogues (TODs), emotions are closely related to task progression. They usually become apparent as the task unfolds, and are contingent on the user's expectations being met [5]. Detecting these emotions explicitly in either scenario is beneficial for several reasons: it can help with reviewing chat logs after the exchange concludes or with adjusting responses during the exchange to better align with the user's emotional state [18]. In TODs, these more empathetic responses have demonstrated their effectiveness in compensating for system errors [12], creating the impression of a more capable system. Existing text-based approaches to ED in TODs either require a dedicated, specifically trained component [5, 6] or assume implicit ED. This assumption is due to systems being commonly trained to replicate human expert responses [25, 7] which inherently convey empathy, as needed, and therefore implicit ED. In contrast, we propose an approach that eliminates the need for training an additional component while explicitly modeling user emotions, applicable in settings where emotion annotations are available.

While several tools have recently been developed for structuring the generation of prompts and collecting responses [4, 41, 45], relatively little effort has been expended to help either end-users or designers of LLM-backed systems to reason about or make use of the variation seen in the multiple responses generated. We see utility in helping users make sense of multiple LLM responses. For instance, users may want to select the best option from among many, compose their own response through bricolage, consider many ideas during ideation, audit a model by looking at the variety of possible responses, or compare the functionality of different models or prompts. However, representing LLM responses at a scale necessary to see the distribution of possibilities also creates a condition where relevant variation may be hidden in plain sight: within a wall of similar text. One could turn to automatic analysis measures, but we constrain ourselves to showing the entirety of the text itself, as this does not constrict (by top-down design or button-up computation) which variations will be most useful to the user.

Autonomous agents capable of planning, reasoning, and executing actions on the web offer a promising avenue for automating computer tasks. However, the majority of existing benchmarks primarily focus on text-based agents, neglecting many natural tasks that require visual information to effectively solve. Given that most computer interfaces cater to human perception, visual information often augments textual data in ways that text-only models struggle to harness effectively. To bridge this gap, we introduce VisualWebArena, a benchmark designed to assess the performance of multimodal web agents on realistic \textit{visually grounded tasks}. VisualWebArena comprises of a set of diverse and complex web-based tasks that evaluate various capabilities of autonomous multimodal agents. To perform on this benchmark, agents need to accurately process image-text inputs, interpret natural language instructions, and execute actions on websites to accomplish user-defined objectives. We conduct an extensive evaluation of state-of-the-art LLM-based autonomous agents, including several multimodal models. Through extensive quantitative and qualitative analysis, we identify several limitations of text-only LLM agents, and reveal gaps in the capabilities of state-of-the-art multimodal language agents. VisualWebArena provides a framework for evaluating multimodal autonomous language agents, and offers insights towards building stronger autonomous agents for the web. Our code, baseline models, and data is publicly available at https://jykoh.com/vwa.

Large Language Models (LLMs) such as GPT developed by OpenAI, have already shown astonishing results, introducing quick changes in our society. This has been intensified by the release of ChatGPT which allows anyone to interact in a simple conversational way with LLMs, without any experience in the field needed. As a result, ChatGPT has been rapidly applied to many different tasks such as code- and song-writer, education, virtual assistants, etc., showing impressive results for tasks for which it was not trained (zero-shot learning). The present study aims to explore the ability of ChatGPT, based on the recent GPT-4 multimodal LLM, for the task of face biometrics. In particular, we analyze the ability of ChatGPT to perform tasks such as face verification, soft-biometrics estimation, and explainability of the results. ChatGPT could be very valuable to further increase the explainability and transparency of the automatic decisions in human scenarios. Experiments are carried out in order to evaluate the performance and robustness of ChatGPT, using popular public benchmarks and comparing the results with state-of-the-art methods in the field. The results achieved in this study show the potential of LLMs such as ChatGPT for face biometrics, especially to enhance explainability. For reproducibility reasons, we release all the code in GitHub.

In the past year, MultiModal Large Language Models (MM-LLMs) have undergone substantial advancements, augmenting off-the-shelf LLMs to support MM inputs or outputs via cost-effective training strategies. The resulting models not only preserve the inherent reasoning and decision-making capabilities of LLMs but also empower a diverse range of MM tasks. In this paper, we provide a comprehensive survey aimed at facilitating further research of MM-LLMs. Specifically, we first outline general design formulations for model architecture and training pipeline. Subsequently, we provide brief introductions of $26$ existing MM-LLMs, each characterized by its specific formulations. Additionally, we review the performance of MM-LLMs on mainstream benchmarks and summarize key training recipes to enhance the potency of MM-LLMs. Lastly, we explore promising directions for MM-LLMs while concurrently maintaining a real-time tracking website for the latest developments in the field. We hope that this survey contributes to the ongoing advancement of the MM-LLMs domain.