Does Thought Require Sensory Grounding? Presidential Address delivered under the title "Can a Large Language Model Think?" at the one hundred nineteenth Eastern Division meeting of the American Philosophical Association on January 6, 2023. Does the capacity to think require the capacity to sense? A lively debate on this topic runs throughout the history of philosophy and now animates discussions of artificial intelligence. In favor of a positive answer, Aristotle says, "The soul never thinks without an image." Aquinas says, "There's nothing in the intellect that wasn't previously in the senses." Hume says, "All our simple ideas in their first appearance are derived from simple impressions." With some minimal assumptions, all three of these statements suggest that thinking requires the capacity to sense, or at least requires having had the capacity to sense at some point. Contrasting with these empiricist theses, rationalist philosophers have often denied that thinking requires sensing. Plato holds that we can think about the forms before we have senses and a body. Descartes holds that the pure intellect thinks independently of the senses. Navigating between empiricism and rationalism, Kant discusses the issue extensively ("Thoughts without content are empty"); unsurprisingly, his final views on the matter are complicated. In recent decades, this philosophical debate has become central to debates in artificial intelligence and cognitive science. He and others held that for symbols to have meaning, they must be causally grounded in sensory connections to the environment. To be meaningful, the symbol "RED" must be grounded in seeing red. The symbol "WATER" must be grounded in a sensory connection to water. If we assume that thinking and meaning go together in AI systems, then this amounts to another version of the thesis that thinking requires sensing. In the last few years, discussion of symbol grounding has become especially widespread in the debate over large language models (LLMs) such as the GPT systems. Can large language models think, mean, or understand?

This study aimed to determine if ChatGPT's large language models could match the scoring accuracy of human and machine scores from the ASAP competition. The investigation focused on various prediction models, including linear regression, random forest, gradient boost, and boost. ChatGPT's performance was evaluated against human raters using quadratic weighted kappa (QWK) metrics. Results indicated that while ChatGPT's gradient boost model achieved QWKs close to human raters for some data sets, its overall performance was inconsistent and often lower than human scores. The study highlighted the need for further refinement, particularly in handling biases and ensuring scoring fairness. Despite these challenges, ChatGPT demonstrated potential for scoring efficiency, especially with domain-specific fine-tuning. The study concludes that ChatGPT can complement human scoring but requires additional development to be reliable for high-stakes assessments. Future research should improve model accuracy, address ethical considerations, and explore hybrid models combining ChatGPT with empirical methods.

Vision-language models (VLMs) are intensively used in many downstream tasks, including those requiring assessments of individuals appearing in the images. While VLMs perform well in simple single-person scenarios, in real-world applications, we often face complex situations in which there are persons of different genders doing different activities. We show that in such cases, VLMs are biased towards identifying the individual with the expected gender (according to ingrained gender stereotypes in the model or other forms of sample selection bias) as the performer of the activity. We refer to this bias in associating an activity with the gender of its actual performer in an image or text as the Gender-Activity Binding (GAB) bias and analyze how this bias is internalized in VLMs. To assess this bias, we have introduced the GAB dataset with approximately 5500 AI-generated images that represent a variety of activities, addressing the scarcity of real-world images for some scenarios. To have extensive quality control, the generated images are evaluated for their diversity, quality, and realism. We have tested 12 renowned pre-trained VLMs on this dataset in the context of text-to-image and image-to-text retrieval to measure the effect of this bias on their predictions. Additionally, we have carried out supplementary experiments to quantify the bias in VLMs' text encoders and to evaluate VLMs' capability to recognize activities. Our experiments indicate that VLMs experience an average performance decline of about 13.2% when confronted with gender-activity binding bias.

The previous advancements in pathology image understanding primarily involved developing models tailored to specific tasks. Recent studies has demonstrated that the large vision-language model can enhance the performance of various downstream tasks in medical image understanding. In this study, we developed a domain-specific large language-vision assistant (PA-LLaVA) for pathology image understanding. Specifically, (1) we first construct a human pathology image-text dataset by cleaning the public medical image-text data for domain-specific alignment; (2) Using the proposed image-text data, we first train a pathology language-image pretraining (PLIP) model as the specialized visual encoder for pathology image, and then we developed scale-invariant connector to avoid the information loss caused by image scaling; (3) We adopt two-stage learning to train PA-LLaVA, first stage for domain alignment, and second stage for end to end visual question \& answering (VQA) task. In experiments, we evaluate our PA-LLaVA on both supervised and zero-shot VQA datasets, our model achieved the best overall performance among multimodal models of similar scale. The ablation experiments also confirmed the effectiveness of our design. We posit that our PA-LLaVA model and the datasets presented in this work can promote research in field of computational pathology. All codes are available at: https://github.com/ddw2AIGROUP2CQUPT/PA-LLaVA}{https://github.com/ddw2AIGROUP2CQUPT/PA-LLaVA

Large language models (LLMs) such as GPT-4 sometimes appear to be creative, solving novel tasks often with a few demonstrations in the prompt. These tasks require the models to generalize on distributions different from those from training data -- which is known as out-of-distribution (OOD) generalization. Despite the tremendous success of LLMs, how they approach OOD generalization remains an open and underexplored question. We examine OOD generalization in settings where instances are generated according to hidden rules, including in-context learning with symbolic reasoning. Models are required to infer the hidden rules behind input prompts without any fine-tuning. We empirically examined the training dynamics of Transformers on a synthetic example and conducted extensive experiments on a variety of pretrained LLMs, focusing on a type of components known as induction heads. We found that OOD generalization and composition are tied together -- models can learn rules by composing two self-attention layers, thereby achieving OOD generalization. Furthermore, a shared latent subspace in the embedding (or feature) space acts as a bridge for composition by aligning early layers and later layers, which we refer to as the common bridge representation hypothesis.

Generative language models (LMs) offer numerous advantages but may produce inappropriate or harmful outputs due to the harmful knowledge acquired during pre-training. This knowledge often manifests as undesirable correspondences, such as "harmful prompts" leading to "harmful outputs," which our research aims to mitigate through unlearning techniques.However, existing unlearning methods based on gradient ascent can significantly impair the performance of LMs. To address this issue, we propose a novel approach called Weighted Positional N-pair (WPN) Learning, which leverages position-weighted mean pooling within an n-pair contrastive learning framework. WPN is designed to modify the output distribution of LMs by eliminating specific harmful outputs (e.g., replacing toxic responses with neutral ones), thereby transforming the model's behavior from "harmful prompt-harmful output" to "harmful prompt-harmless response".Experiments on OPT and GPT-NEO LMs show that WPN effectively reduces the proportion of harmful responses, achieving a harmless rate of up to 95.8\% while maintaining stable performance on nine common benchmarks (with less than 2\% degradation on average). Moreover, we provide empirical evidence to demonstrate WPN's ability to weaken the harmful correspondences in terms of generalizability and robustness, as evaluated on out-of-distribution test sets and under adversarial attacks.

Quality assessment, which evaluates the visual quality level of multimedia experiences, has garnered significant attention from researchers and has evolved substantially through dedicated efforts. Before the advent of large models, quality assessment typically relied on small expert models tailored for specific tasks. While these smaller models are effective at handling their designated tasks and predicting quality levels, they often lack explainability and robustness. With the advancement of large models, which align more closely with human cognitive and perceptual processes, many researchers are now leveraging the prior knowledge embedded in these large models for quality assessment tasks. This emergence of quality assessment within the context of large models motivates us to provide a comprehensive review focusing on two key aspects: 1) the assessment of large models, and 2) the role of large models in assessment tasks. We begin by reflecting on the historical development of quality assessment. Subsequently, we move to detailed discussions of related works concerning quality assessment in the era of large models. Finally, we offer insights into the future progression and potential pathways for quality assessment in this new era. We hope this survey will enable a rapid understanding of the development of quality assessment in the era of large models and inspire further advancements in the field.

This paper explores the application of Stochastic Differential Equations (SDE) to interpret the text generation process of Large Language Models (LLMs) such as GPT-4. Text generation in LLMs is modeled as a stochastic process where each step depends on previously generated content and model parameters, sampling the next word from a vocabulary distribution. We represent this generation process using SDE to capture both deterministic trends and stochastic perturbations. The drift term describes the deterministic trends in the generation process, while the diffusion term captures the stochastic variations. We fit these functions using neural networks and validate the model on real-world text corpora. Through numerical simulations and comprehensive analyses, including drift and diffusion analysis, stochastic process property evaluation, and phase space exploration, we provide deep insights into the dynamics of text generation. This approach not only enhances the understanding of the inner workings of LLMs but also offers a novel mathematical perspective on language generation, which is crucial for diagnosing, optimizing, and controlling the quality of generated text.

Large Language Models (LLMs) are highly sensitive to prompts, including additional context provided therein. As LLMs grow in capability, understanding their prompt-sensitivity becomes increasingly crucial for ensuring reliable and robust performance, particularly since evaluating these models becomes more challenging. In this work, we investigate how current models (Llama, Mixtral, Falcon) respond when presented with additional input from another model, mimicking a scenario where a more capable model -- or a system with access to more external information -- provides supplementary information to the target model. Across a diverse spectrum of question-answering tasks, we study how an LLM's response to multiple-choice questions changes when the prompt includes a prediction and explanation from another model. Specifically, we explore the influence of the presence of an explanation, the stated authoritativeness of the source, and the stated confidence of the supplementary input. Our findings reveal that models are strongly influenced, and when explanations are provided they are swayed irrespective of the quality of the explanation. The models are more likely to be swayed if the input is presented as being authoritative or confident, but the effect is small in size. This study underscores the significant prompt-sensitivity of LLMs and highlights the potential risks of incorporating outputs from external sources without thorough scrutiny and further validation. As LLMs continue to advance, understanding and mitigating such sensitivities will be crucial for their reliable and trustworthy deployment.

This study introduces the Federated Medical Knowledge Injection (FEDMEKI) platform, a new benchmark designed to address the unique challenges of integrating medical knowledge into foundation models under privacy constraints. By leveraging a cross-silo federated learning approach, FEDMEKI circumvents the issues associated with centralized data collection, which is often prohibited under health regulations like the Health Insurance Portability and Accountability Act (HIPAA) in the USA. The platform is meticulously designed to handle multi-site, multi-modal, and multi-task medical data, which includes 7 medical modalities, including images, signals, texts, laboratory test results, vital signs, input variables, and output variables. The curated dataset to validate FEDMEKI covers 8 medical tasks, including 6 classification tasks (lung opacity detection, COVID-19 detection, electrocardiogram (ECG) abnormal detection, mortality prediction, sepsis prediction, and enlarged cardiomediastinum detection) and 2 generation tasks (medical visual question answering (MedVQA) and ECG noise clarification). This comprehensive dataset is partitioned across several clients to facilitate the decentralized training process under 16 benchmark approaches. FEDMEKI not only preserves data privacy but also enhances the capability of medical foundation models by allowing them to learn from a broader spectrum of medical knowledge without direct data exposure, thereby setting a new benchmark in the application of foundation models within the healthcare sector.