Machine psychology aims to reconstruct the mindset of Large Language Models (LLMs), i.e. how these artificial intelligences perceive and associate ideas. This work introduces PhDGPT, a prompting framework and synthetic dataset that encapsulates the machine psychology of PhD researchers and professors as perceived by OpenAI's GPT-3.5. The dataset consists of 756,000 datapoints, counting 300 iterations repeated across 15 academic events, 2 biological genders, 2 career levels and 42 unique item responses of the Depression, Anxiety, and Stress Scale (DASS-42). PhDGPT integrates these psychometric scores with their explanations in plain language. This synergy of scores and texts offers a dual, comprehensive perspective on the emotional well-being of simulated academics, e.g. male/female PhD students or professors. By combining network psychometrics and psycholinguistic dimensions, this study identifies several similarities and distinctions between human and LLM data. The psychometric networks of simulated male professors do not differ between physical and emotional anxiety subscales, unlike humans. Other LLMs' personification can reconstruct human DASS factors with a purity up to 80%. Furthemore, LLM-generated personifications across different scenarios are found to elicit explanations lower in concreteness and imageability in items coding for anxiety, in agreement with past studies about human psychology. Our findings indicate an advanced yet incomplete ability for LLMs to reproduce the complexity of human psychometric data, unveiling convenient advantages and limitations in using LLMs to replace human participants. PhDGPT also intriguingly capture the ability for LLMs to adapt and change language patterns according to prompted mental distress contextual features, opening new quantitative opportunities for assessing the machine psychology of these artificial intelligences.

Enterprise searches require users to have complex knowledge of queries, configurations, and metadata, rendering it difficult for them to access information as needed. Most go-to-market (GTM) platforms utilize advanced search, an interface that enables users to filter queries by various fields using categories or keywords, which, historically, however, has proven to be exceedingly cumbersome, as users are faced with seemingly hundreds of options, fields, and buttons. Consequently, querying with natural language has long been ideal, a notion further empowered by Large Language Models (LLMs). In this paper, we implement and evaluate a solution for the Zoominfo product for sellers, which prompts the LLM with natural language, producing search fields through entity extraction that are then converted into a search query. The intermediary search fields offer numerous advantages for each query, including the elimination of syntax errors, simpler ground truths, and an intuitive format for the LLM to interpret. We paired this pipeline with many advanced prompt engineering strategies, featuring an intricate system message, few-shot prompting, chain-of-thought (CoT) reasoning, and execution refinement. Furthermore, we manually created the ground truth for 500+ natural language queries, enabling the supervised fine-tuning of Llama-3-8B-Instruct and the introduction of sophisticated numerical metrics. Comprehensive experiments with closed, open source, and fine-tuned LLM models were conducted through exact, Jaccard, cosine, and semantic similarity on individual search entities to demonstrate the efficacy of our approach. Overall, the most accurate closed model had an average accuracy of 97% per query, with only one field performing under 90%, with comparable results observed from the fine-tuned models.

In this study, we aim to enhance radiology reporting by improving both the conciseness and structured organization of findings (also referred to as templating), specifically by organizing information according to anatomical regions. This structured approach allows physicians to locate relevant information quickly, increasing the report's utility. We utilize Large Language Models (LLMs) such as Mixtral, Mistral, and Llama to generate concise, well-structured reports. Among these, we primarily focus on the Mixtral model due to its superior adherence to specific formatting requirements compared to other models. To maintain data security and privacy, we run these LLMs locally behind our institution's firewall. We leverage the LangChain framework and apply five distinct prompting strategies to enforce a consistent structure in radiology reports, aiming to eliminate extraneous language and achieve a high level of conciseness. We also introduce a novel metric, the Conciseness Percentage (CP) score, to evaluate report brevity. Our dataset comprises 814 radiology reports authored by seven board-certified body radiologists at our cancer center. In evaluating the different prompting methods, we discovered that the most effective approach for generating concise, well-structured reports involves first instructing the LLM to condense the report, followed by a prompt to structure the content according to specific guidelines. We assessed all prompting strategies based on their ability to handle formatting issues, reduce report length, and adhere to formatting instructions. Our findings demonstrate that open-source, locally deployed LLMs can significantly improve radiology report conciseness and structure while conforming to specified formatting standards.

Short video platforms, such as YouTube, Instagram, or TikTok, are used by billions of users globally. These platforms expose users to harmful content, ranging from clickbait or physical harms to misinformation or online hate. Yet, detecting harmful videos remains challenging due to an inconsistent understanding of what constitutes harm and limited resources and mental tolls involved in human annotation. As such, this study advances measures and methods to detect harm in video content. First, we develop a comprehensive taxonomy for online harm on video platforms, categorizing it into six categories: Information, Hate and harassment, Addictive, Clickbait, Sexual, and Physical harms. Next, we establish multimodal large language models as reliable annotators of harmful videos. We analyze 19,422 YouTube videos using 14 image frames, 1 thumbnail, and text metadata, comparing the accuracy of crowdworkers (Mturk) and GPT-4-Turbo with domain expert annotations serving as the gold standard. Our results demonstrate that GPT-4-Turbo outperforms crowdworkers in both binary classification (harmful vs. harmless) and multi-label harm categorization tasks. Methodologically, this study extends the application of LLMs to multi-label and multi-modal contexts beyond text annotation and binary classification. Practically, our study contributes to online harm mitigation by guiding the definitions and identification of harmful content on video platforms.

Embeddings have become a cornerstone in the functionality of large language models (LLMs) due to their ability to transform text data into rich, dense numerical representations that capture semantic and syntactic properties. These embedding vector databases serve as the long-term memory of LLMs, enabling efficient handling of a wide range of natural language processing tasks. However, the surge in popularity of embedding vector databases in LLMs has been accompanied by significant concerns about privacy leakage. Embedding vector databases are particularly vulnerable to embedding inversion attacks, where adversaries can exploit the embeddings to reverse-engineer and extract sensitive information from the original text data. Existing defense mechanisms have shown limitations, often struggling to balance security with the performance of downstream tasks. To address these challenges, we introduce Eguard, a novel defense mechanism designed to mitigate embedding inversion attacks. Eguard employs a transformer-based projection network and text mutual information optimization to safeguard embeddings while preserving the utility of LLMs. Our approach significantly reduces privacy risks, protecting over 95% of tokens from inversion while maintaining high performance across downstream tasks consistent with original embeddings.

Geometric Dimensioning and Tolerancing (GD&T) plays a critical role in manufacturing by defining acceptable variations in part features to ensure component quality and functionality. However, extracting GD&T information from 2D engineering drawings is a time-consuming and labor-intensive task, often relying on manual efforts or semi-automated tools. To address these challenges, this study proposes an automated and computationally efficient GD&T extraction method by fine-tuning Florence-2, an open-source vision-language model (VLM). The model is trained on a dataset of 400 drawings with ground truth annotations provided by domain experts. For comparison, two state-of-the-art closed-source VLMs, GPT-4o and Claude-3.5-Sonnet, are evaluated on the same dataset. All models are assessed using precision, recall, F1-score, and hallucination metrics. Due to the computational cost and impracticality of fine-tuning large closed-source VLMs for domain-specific tasks, GPT-4o and Claude-3.5-Sonnet are evaluated in a zero-shot setting. In contrast, Florence-2, a smaller model with 0.23 billion parameters, is optimized through full-parameter fine-tuning across three distinct experiments, each utilizing datasets augmented to different levels. The results show that Florence-2 achieves a 29.95% increase in precision, a 37.75% increase in recall, a 52.40% improvement in F1-score, and a 43.15% reduction in hallucination rate compared to the best-performing closed-source model. These findings highlight the effectiveness of fine-tuning smaller, open-source VLMs like Florence-2, offering a practical and efficient solution for automated GD&T extraction to support downstream manufacturing tasks.

We use images of cars of a wide range of varieties to compose an image of an animal such as a bird or a lion for the theme of environmental protection to maximize the information about cars in a single composed image and to raise the awareness about environmental challenges. We present a novel way of image interaction with an artistically-composed photomosaic image, in which a simple operation of "click and display" is used to demonstrate the interactive switch between a tile image in a photomosaic image and the corresponding original car image, which will be automatically saved on the Desktop. We build a multimodal custom GPT named TalkMosaic by incorporating car images information and the related knowledge to ChatGPT. By uploading the original car image to TalkMosaic, we can ask questions about the given car image and get the corresponding answers efficiently and effectively such as where to buy the tire in the car image that satisfies high environmental standards. We give an in-depth analysis on how to speed up the inference of multimodal LLM using sparse attention and quantization techniques with presented probabilistic FlashAttention (PrFlashAttention) and Staircase Adaptive Quantization (SAQ) methods. The implemented prototype demonstrates the feasibility and effectiveness of the presented approach.

With the growing complexity and capability of large language models, a need to understand model reasoning has emerged, often motivated by an underlying goal of controlling and aligning models. While numerous interpretability and steering methods have been proposed as solutions, they are typically designed either for understanding or for control, seldom addressing both, with the connection between interpretation and control more broadly remaining tenuous. Additionally, the lack of standardized applications, motivations, and evaluation metrics makes it difficult to assess these methods' practical utility and efficacy. To address this, we propose intervention as a fundamental goal of interpretability and introduce success criteria to evaluate how well methods are able to control model behavior through interventions. We unify and extend four popular interpretability methods--sparse autoencoders, logit lens, tuned lens, and probing--into an abstract encoder-decoder framework. This framework maps intermediate latent representations to human-interpretable feature spaces, enabling interventions on these interpretable features, which can then be mapped back to latent representations to control model outputs. We introduce two new evaluation metrics: intervention success rate and the coherence-intervention tradeoff, designed to measure the accuracy of explanations and their utility in controlling model behavior. Our findings reveal that (1) although current methods allow for intervention, they are inconsistent across models and features, (2) lens-based methods outperform others in achieving simple, concrete interventions, and (3) interventions often compromise model performance and coherence, underperforming simpler alternatives, such as prompting, for steering model behavior and highlighting a critical shortcoming of current interpretability approaches in real-world applications requiring control.

However, although the cooperation between these two seems natural, the difference in their representation is obviously not negligible. Prof. Henry Kautz proposed a taxonomy of Neuro-Symbolic Systems in the AAAI 2020. In addition, many researchers have conducted relevant reviews of the recent neuro-symbolic AI from different perspectives. As Fig.1 shows, Acharya et al. [1] proposed a new classification method, which classified and discussed the application of existing neuro-symbolic AI by the role of neural and symbolic parts: learning for reasoning, reasoning for Learning, and learning-reasoning. Garcez et al. [73] proposed a taxonomy that includes sequential, nested, cooperative, and compiled neuro-symbolic AI based on the six types introduced by Henry Kautz. In addition, some reviews focus on cross-field integration and applications. For example, Berlot-Attwell [27] reviewed neuro-symbolic VQA (visual question answering) from the perspectives of AGI (artificial general intelligence) desiderata. Marra [128] conducted a comprehensive review on integrating neuro-symbolic and statistical relational artificial intelligence based on seven dimensions.

Machine unlearning aims to solve the problem of removing the influence of selected training examples from a learned model. Despite the increasing attention to this problem, it remains an open research question how to evaluate unlearning in large language models (LLMs), and what are the critical properties of the data to be unlearned that affect the quality and efficiency of unlearning. This work formalizes a metric to evaluate unlearning quality in generative models, and uses it to assess the trade-offs between unlearning quality and performance. We demonstrate that unlearning out-of-distribution examples requires more unlearning steps but overall presents a better trade-off overall. For in-distribution examples, however, we observe a rapid decay in performance as unlearning progresses. We further evaluate how example's memorization and difficulty affect unlearning under a classical gradient ascent-based approach.