Language Models Surface the Unwritten Code of Science and Society

This paper calls on the research community not only to investigate how human biases are inherited by large language models (LLMs) but also to explore how these biases in LLMs can be leveraged to make society's "unwritten code" - such as implicit stereotypes and heuristics - visible and accessible for critique. We introduce a conceptual framework through a case study in science: uncovering hidden rules in peer review - the factors that reviewers care about but rarely state explicitly due to normative scientific expectations. The idea of the framework is to push LLMs to speak out their heuristics through generating self-consistent hypotheses - why one paper appeared stronger in reviewer scoring - among paired papers submitted to 45 computer science conferences, while iteratively searching deeper hypotheses from remaining pairs where existing hypotheses cannot explain. We observed that LLMs' normative priors about the internal characteristics of good science extracted from their self-talk, e.g. theoretical rigor, were systematically updated toward posteriors that emphasize storytelling about external connections, such as how the work is positioned and connected within and across literatures. This shift reveals the primacy of scientific myths about intrinsic properties driving scientific excellence rather than extrinsic contextualization and storytelling that influence conceptions of relevance and significance. Human reviewers tend to explicitly reward aspects that moderately align with LLMs' normative priors (correlation = 0.49) but avoid articulating contextualization and storytelling posteriors in their review comments (correlation = -0.14), despite giving implicit reward to them with positive scores. We discuss the broad applicability of the framework, leveraging LLMs as diagnostic tools to surface the tacit codes underlying human society, enabling more precisely targeted responsible AI.

Introspective Growth: Automatically Advancing LLM Expertise in Technology Judgment

Large language models (LLMs) increasingly demonstrate signs of conceptual understanding, yet much of their internal knowledge remains latent, loosely structured, and difficult to access or evaluate. We propose self-questioning as a lightweight and scalable strategy to improve LLMs' understanding, particularly in domains where success depends on fine-grained semantic distinctions. To evaluate this approach, we introduce a challenging new benchmark of 1.3 million post-2015 computer science patent pairs, characterized by dense technical jargon and strategically complex writing. The benchmark centers on a pairwise differentiation task: can a model distinguish between closely related but substantively different inventions? We show that prompting LLMs to generate and answer their own questions - targeting the background knowledge required for the task - significantly improves performance. These self-generated questions and answers activate otherwise underutilized internal knowledge. Allowing LLMs to retrieve answers from external scientific texts further enhances performance, suggesting that model knowledge is compressed and lacks the full richness of the training data. We also find that chain-of-thought prompting and self-questioning converge, though self-questioning remains more effective for improving understanding of technical concepts. Notably, we uncover an asymmetry in prompting: smaller models often generate more fundamental, more open-ended, better-aligned questions for mid-sized models than large models with better understanding do, revealing a new strategy for cross-model collaboration. Altogether, our findings establish self-questioning as both a practical mechanism for automatically improving LLM comprehension, especially in domains with sparse and underrepresented knowledge, and a diagnostic probe of how internal and external knowledge are organized.

China and the U.S. produce more impactful AI research when collaborating together

Artificial Intelligence (AI) has become a disruptive technology, promising to grant a significant economic and strategic advantage to the nations that harness its power. China, with its recent push towards AI adoption, is challenging the U.S.'s position as the global leader in this field. Given AI's massive potential, as well as the fierce geopolitical tensions between the two nations, a number of policies have been put in place that discourage AI scientists from migrating to, or collaborating with, the other country. However, the extents of such brain drain and cross-border collaboration are not fully understood. Here, we analyze a dataset of over 350,000 AI scientists and 5,000,000 AI papers. We find that, since the year 2000, China and the U.S. have been leading the field in terms of impact, novelty, productivity, and workforce. Most AI scientists who migrate to China come from the U.S., and most who migrate to the U.S. come from China, highlighting a notable brain drain in both directions. Upon migrating from one country to the other, scientists continue to collaborate frequently with the origin country. Although the number of collaborations between the two countries has been increasing since the dawn of the millennium, such collaborations continue to be relatively rare. A matching experiment reveals that the two countries have always been more impactful when collaborating than when each of them works without the other. These findings suggest that instead of suppressing cross-border migration and collaboration between the two nations, the field could benefit from promoting such activities.

Learning from One and Only One Shot

Humans can generalize from only a few examples and from little pre-training on similar tasks. Yet, machine learning (ML) typically requires large data to learn or pre-learn to transfer. Inspired by nativism, we directly model basic human-innate priors in abstract visual tasks e.g., character/doodle recognition. This yields a white-box model that learns general-appearance similarity -- how any two images look in general -- by mimicking how humans naturally "distort" an object at first sight. Using simply the nearest-neighbor classifier on this similarity space, we achieve human-level character recognition using only 1--10 examples per class and nothing else (no pre-training). This differs from few-shot learning (FSL) using significant pre-training. On standard benchmarks MNIST/EMNIST and the Omniglot challenge, we outperform both neural-network-based and classical ML in the "tiny-data" regime, including FSL pre-trained on large data. Our model enables unsupervised learning too: by learning the non-Euclidean, general-appearance similarity space in a k-means style, we can generate human-intuitive archetypes as cluster ``centroids''.

Human Languages with Greater Information Density Increase Communication Speed, but Decrease Conversation Breadth

Language is the primary medium through which human information is communicated and coordination is achieved. One of the most important language functions is to categorize the world so messages can be communicated through conversation. While we know a great deal about how human languages vary in their encoding of information within semantic domains such as color, sound, number, locomotion, time, space, human activities, gender, body parts and biology, little is known about the global structure of semantic information and its effect on human communication. Using large-scale computation, artificial intelligence techniques, and massive, parallel corpora across 15 subject areas--including religion, economics, medicine, entertainment, politics, and technology--in 999 languages, here we show substantial variation in the information and semantic density of languages and their consequences for human communication and coordination. In contrast to prior work, we demonstrate that higher density languages communicate information much more quickly relative to lower density languages. Then, using over 9,000 real-life conversations across 14 languages and 90,000 Wikipedia articles across 140 languages, we show that because there are more ways to discuss any given topic in denser languages, conversations and articles retrace and cycle over a narrower conceptual terrain. These results demonstrate an important source of variation across the human communicative channel, suggesting that the structure of language shapes the nature and texture of conversation, with important consequences for the behavior of groups, organizations, markets, and societies.

Human Evaluation of Interpretability: The Case of AI-Generated Music Knowledge

Interpretability of machine learning models has gained more and more attention among researchers in the artificial intelligence (AI) and human-computer interaction (HCI) communities. Most existing work focuses on decision making, whereas we consider knowledge discovery. In particular, we focus on evaluating AI-discovered knowledge/rules in the arts and humanities. From a specific scenario, we present an experimental procedure to collect and assess human-generated verbal interpretations of AI-generated music theory/rules rendered as sophisticated symbolic/numeric objects. Our goal is to reveal both the possibilities and the challenges in such a process of decoding expressive messages from AI sources. We treat this as a first step towards 1) better design of AI representations that are human interpretable and 2) a general methodology to evaluate interpretability of AI-discovered knowledge representations.

Too many cooks: Coordinating multi-agent collaboration through inverse planning

Collaboration requires agents to coordinate their behavior on the fly, sometimes cooperating to solve a single task together and other times dividing it up into sub-tasks to work on in parallel. Underlying the human ability to collaborate is theory-of-mind, the ability to infer the hidden mental states that drive others to act. Here, we develop Bayesian Delegation, a decentralized multi-agent learning mechanism with these abilities. Bayesian Delegation enables agents to rapidly infer the hidden intentions of others by inverse planning. These inferences enable agents to flexibly decide in the absence of communication when to cooperate on the same sub-task and when to work on different sub-tasks in parallel. We test this model in a suite of multi-agent Markov decision processes inspired by cooking problems. To succeed, agents must coordinate both their high-level plans (e.g., what sub-task they should work on) and their low-level actions (e.g., avoiding collisions). Bayesian Delegation bridges these two levels and rapidly aligns agents' beliefs about who should work on what without any communication. When agents cooperate on the same sub-task, coordinated plans emerge that enable the group of agents to achieve tasks no agent can complete on their own. Our model outperforms lesioned agents without Bayesian Delegation or without the ability to cooperate on the same sub-task.

The Geometry of Culture: Analyzing Meaning through Word Embeddings

We demonstrate the utility of a new methodological tool, neural-network word embedding models, for large-scale text analysis, revealing how these models produce richer insights into cultural associations and categories than possible with prior methods. Word embeddings represent semantic relations between words as geometric relationships between vectors in a high-dimensional space, operationalizing a relational model of meaning consistent with contemporary theories of identity and culture. We show that dimensions induced by word differences (e.g. man - woman, rich - poor, black - white, liberal - conservative) in these vector spaces closely correspond to dimensions of cultural meaning, and the projection of words onto these dimensions reflects widely shared cultural connotations when compared to surveyed responses and labeled historical data. We pilot a method for testing the stability of these associations, then demonstrate applications of word embeddings for macro-cultural investigation with a longitudinal analysis of the coevolution of gender and class associations in the United States over the 20th century and a comparative analysis of historic distinctions between markers of gender and class in the U.S. and Britain. We argue that the success of these high-dimensional models motivates a move towards "high-dimensional theorizing" of meanings, identities and cultural processes.