arXiv subjects

James A. Evans

Publications and source records attributed to James A. Evans.

At least 19 recordsLinked to original sources

arXiv2026

Narrative Flattening: How Post-Training Compresses Thematic, Affective, and Stylistic Variation in LLM Fiction

Large language models produce fluent fiction, yet their creative output is widely seen as flat. We ask where this quality originates in the training and whether it affects different domains of human fiction equally. We construct a matched story-continuation paradigm across StoryStar (public-platform), TMAS (prompt-guided), and The New Yorker (professional literary)-and compare continuations from four OLMo 32B checkpoints (Base, SFT, DPO, RLVR) against matched human text. Because these checkpoints share architecture, scale, tokenizer, and pretraining, the design isolates the post-training effect. We measure each continuation along three sentence-level dimensions: thematic motion, affective prevalence, and linguistic diversity. Across all three, post-training compresses dynamic variation: thematic transitions become more uniform, high-intensity emotions give way to neutrality, and stylistic diversity across stories shrinks. We term this progressive loss narrative flattening. The effect is directionally stable across story domains but gap size depends on the human baseline: professional literary fiction is compressed most, while public-platform and prompt-guided stories show smaller gaps, consistent with their human baselines sitting closer to the model's default rhythm. Post-trained endpoints converge across domains, suggesting alignment produces a continuation regime largely insensitive to the source domain's narrative texture.

cs.CL↗

arXiv2026

The Illusion of Debiasing: Persona Steering Redistributes Rather Than Reduces Bias in LLMs

Prompt-based interventions: system prompts, personas, role instructions, reliably reshape what a language model says, but it is unclear which layer they reach. Do they reconfigure internal structure, or only modulate the output channel? We use persona conditioning as a controlled probe, measuring its effects along a depth axis from self-report, through open-ended generation, to word-level parametric association, across three instruction-tuned models. We find a graded dissociation. Personas are legible but not structural: models follow single-trait instructions yet fail to reproduce human inter-trait covariance. The dissociation deepens with depth: personas hold or amplify closed-form QA bias, shift absolute tone while leaving between-group disparity unchanged, and barely perturb an already saturated associative baseline. Prompt-based steering thus operates in the output channel and has a structural reach limit that surface manipulability can mask.

cs.CL↗

arXiv2026

Filling holes in science draws collective attention, but most higher-order holes remain unexplored

Much scientific discovery involves filling holes between ideas and arguments that unleash techno-scientific advance. Representing knowledge as high-dimensional concept embeddings, we use persistent homology to detect holes of increasing order, from gaps between disconnected ideas to higher-order cavities, and identify the research works that fill them. We find two empirical asymmetries. Researchers who fill anticipated holes are poised to draw collective attention by staging outsized novelty and foresight, indicating that bridging holes anticipates where science will converge, most strongly in empirical fields and least in formal and design fields. Yet as knowledge grows, higher-order holes explode while the fraction science fills collapses, leaving most higher-order combinations unexplored. These results call for a richer science of holes, and mark a frontier where contemporary AI might help fill the high-dimensional gaps human science opens.

cs.CY↗

arXiv2026

Contemporary AI lacks the imagination to diverge or negate in science

Bold claims that AI will accelerate scientific discovery have raced ahead of evidence from working scientists, yet large-scale, scientist-in-the-loop evidence is scarce. Here we mount the largest evaluation to date, inviting authors of 121,640 recent preprints in biology, medicine, chemistry, and social science to judge large language model (LLM)-generated ideas derived from their own papers. 6,749 representative scientists returned 25,139 rating sets on novelty, feasibility, probability of being true, and favorability of adoption. Three patterns emerge. First, non-reasoning LLMs collapse into a narrow "hivemind" of similar ideas while reasoning models explore a wider hypothesis space, but no model spontaneously proposes null hypotheses, a move humans make more freely. Second, scientists reward ideas resembling their own and prize probability over novelty, though social scientists tolerate risk more than life scientists; senior social scientists are the harshest critics, and their skepticism is earned, as LLMs falter most in pluralistic fields demanding context-aware interpretation and evolving theories. Third, automated evaluators, including LLM-as-a-judge and state-of-the-art (SOTA) models, agree weakly with expert judgment. Retrieval augmentation and scientist persona prompting yield marginal gains. A Qwen3-14B reward model we post-trained on human ratings captures nuances of taste, beats SOTA models by up to 27%, and closes the gap to the consistency of human peer reviewers. An analysis of 39 million papers from 2010 to 2025 links survey findings to macro-level patterns: following ChatGPT's release, null claims are sharply suppressed and ideas contract. Agent-based simulations further suggest that saturated fields should especially prize human uniqueness. For all the hype, today's AI for science remains a collaborator whose imagination and judgment benefit from human grounding.

cs.CY↗

arXiv2026

Measuring Intent Comprehension in LLMs

People judge interactions with large language models (LLMs) as successful when outputs match what they want, not what they type. Yet LLMs are trained to predict the next token solely from text input, not underlying intent. Because written language is an imperfect proxy for intent, and correlations between phrasing and desired outcomes can break down in training data, models that rely too heavily on surface cues may respond inconsistently to semantically equivalent prompts. This makes it essential to evaluate whether LLMs can reliably infer user intent-especially in high-stakes settings where robustness and generalization are critical. We introduce a formal framework for assessing intent comprehension in LLMs: whether a model demonstrates robust understanding of user intent by producing consistent outputs across semantically equivalent prompts while differentiating between prompts with distinct intents. Our evaluation approach is based on a variance decomposition of model responses into three components: variability due to user intent, user articulation, and model uncertainty. Models that understand what users want, and are not overly sensitive to textual cues, should attribute most output variance to intent differences, rather than articulation style. Applying this framework across diverse domains, we find that, within the five LLaMA and Gemma models we evaluate, larger models typically assign a greater share of variance to intent, indicating stronger comprehension of intent, although gains are uneven and often modest with increasing model size. These results motivate moving beyond accuracy-only benchmarks toward semantic diagnostics that directly assess whether models understand what users intend.

cs.CL↗

arXiv2026

Measuring Behavior Portability in Large Language Models

Large language models are increasingly deployed as autonomous decision makers, yet the behavioral mapping they exhibit can vary substantially across decision environments that are payoff-equivalent by construction-environments that share identical payoff-relevant structure but differ in surface presentation. This sensitivity renders suite-based evaluation fragile and raises a fundamental question of behavioral portability: how well does a behavioral mapping learned in one decision environment informative on another that preserves the same underlying incentive structure? We introduce a formal framework to measure this property. Our protocol fits an interpretable behavioral model on data pooled from a set of source environments and evaluates its out-of-sample predictive performance in a held-out target environment, benchmarking against an oracle trained directly on target data. Portability is quantified via a loss-agnostic measure that delivers worst-case bounds on the performance of the induced prediction-action mapping in the target environment. In controlled experiments spanning seven canonical economic decision problems, we document substantial and systematic portability losses, suggesting that behavioral characterizations of LLMs obtained in one decision environment cannot be assumed to transfer reliably to structurally equivalent alternatives.

cs.AI↗

arXiv2026

Building an Atlas of Social Experiments to Link Studies, Reconcile Conflicts, and Bridge Gaps

Social and behavioral science runs thousands of experiments each year, yet their findings rarely accumulate into a coherent map of what is known, what conflicts, and what remains missing. We introduce ExAtlas, a framework for turning an archive of experiments into an atlas: a structured map in which studies link, conflict, or leave bridgeable gaps. Given a target study, ExAtlas searches for prior studies that are locally close in treatment and outcome space and asks whether their observed effects can be composed to predict the target effect. This yields three cases. If the composition succeeds and agrees with the observed result, ExAtlas links the target to consistent prior evidence. If composition succeeds but disagrees, ExAtlas reconciles the conflict and proposes candidate moderators or higher-level theories that could explain it. If composition fails, ExAtlas proposes bridge experiments to close the gap. We provide an error bound for composition under local smoothness of the treatment-effect surface. On held-out targets certified as locally supported, ExAtlas recovers effect direction in 98.6% of cases. Human evaluations further suggest that its proposed bridge experiments are plausible and exhibit connectedness, and that its conflict explanations are useful for theory generation. These results suggest that the archive of social experiments contains more latent structure than current practice extracts -- and that making this structure explicit can guide both future theory and future experimentation.

cs.CY↗

arXiv2026

A Theory of Appropriateness That Accounts for Norms of Rationality

We propose a society-first theory of normative appropriateness where individuals, modeled as pre-trained actors with cognitive architectures analogous to Large Language Models (LLMs), generate behavior via predictive pattern completion. Our theory posits that individuals act by completing distributed symbolic patterns based on context, answering questions such as "What does a person such as I do in a situation such as this?". This sense-making mechanism provides a parsimonious account of the key features of human norms: their context-dependence, arbitrariness, automaticity, dynamism, and their support from social sanctioning. It challenges rational-choice theories of social norms by accounting for their key features without needing to exogenously posit scalar rewards or preference relations. By distinguishing between explicit norms, which we associate with in-context adaptation, and implicit norms, which we associate with long-term memory, the theory reconceptualizes several foundational ideas in cognitive science. In particular, it gives an alternative account to the data traditionally seen as supporting dual-process models, and it flips the role of rationality, allowing us to construe it as adherence to culturally-contingent justification standards.

cs.NE↗

arXiv2026

Mapping Overlaps in Benchmarks through Perplexity in the Wild

We introduce benchmark signatures to characterize the capacity demands of LLM benchmarks and their overlaps. Signatures are sets of salient tokens from in-the-wild corpora whose model token perplexity, reflecting training exposure, predicts benchmark performance. We extract them via stepwise forward selection with linear regression in a meta-evaluation spanning 32 LLMs and 89 benchmarks across diverse domains. We then analyze how these signatures relate to both the semantic similarity of benchmark questions and the correlation structure of model performance. While performance correlations are uniformly high and semantic overlaps stay in a narrow mid-range, benchmark signatures reveal more nuanced structure. For instance, they uncover substantial overlap between benchmarks in knowledge and reasoning tasks, whereas benchmarks in culture- and humanity-oriented domains show low similarity with each other. Unlike raw performance correlations, which are influenced by benchmark-orthogonal factors such as question formats, signatures are robust to such confounds. We further identify cross-functional overlaps between logic, math, language, instruction following, and cultural/world modeling, with coding emerging as the most isolated function, interacting only moderately with the ability of detecting missing information. Qualitative analysis shows that only the knowledge signature aligns with actual knowledge, suggesting that LLM semantic organization may differ from human conceptual structure. Together, these findings offer insights into benchmark validity, LLM sensitivities, and the landscape of interconnected LLM capacities. We have open-sourced the code and data in this https://github.com/siyangwu1/Benchmark-Signature-Repository.

cs.AI↗

arXiv2026

Language Models Should be Used to 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 46 academic 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. 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. These patterns are robust across different models and out-of-sample judgments. We discuss the broad applicability of our proposed framework, leveraging LLMs as diagnostic tools to amplify and surface the tacit codes underlying human society, enabling public discussion of revealed values and more precisely targeted responsible AI.

cs.CY↗

arXiv2026

Missing vs. Unused Knowledge Hypothesis for Language Model Bottlenecks in Patent Understanding

While large language models (LLMs) excel at factual recall, the real challenge lies in knowledge application. A gap persists between their ability to answer complex questions and their effectiveness in performing tasks that require that knowledge. We investigate this gap using a patent classification problem that requires deep conceptual understanding to distinguish semantically similar but objectively different patents written in dense, strategic technical language. We find that LLMs often struggle with this distinction. To diagnose the source of these failures, we introduce a framework that decomposes model errors into two categories: missing knowledge and unused knowledge. Our method prompts models to generate clarifying questions and compares three settings -- raw performance, self-answered questions that activate internal knowledge, and externally provided answers that supply missing knowledge (if any). We show that most errors stem from failures to deploy existing knowledge rather than from true knowledge gaps. We also examine how models differ in constructing task-specific question-answer databases. Smaller models tend to generate simpler questions that they, and other models, can retrieve and use effectively, whereas larger models produce more complex questions that are less effective, suggesting complementary strengths across model scales. Together, our findings highlight that shifting evaluation from static fact recall to dynamic knowledge application offers a more informative view of model capabilities.

cs.CL↗

arXiv2025

Aging and the Narrowing of Scientific Innovation

With rising life expectancies around the world and an older scientific workforce than ever before, what does aging mean for individual scientists, and what do aging scientists mean for scientific progress as a whole? Here we examine how scientists and scholars age in terms of how their ideas and contributions relate to the evolving frontier of knowledge and how demographically aging fields relate to field-level advance. At the individual level, we examine how research experiences and choices can moderate the effects of intellectual aging. At the collective level, we explore mechanisms that link individual and collective aging. Prior research focuses on star scientists, their changing dates and rates of breakthrough success throughout history. We explore this for scientists in all fields over time, drawing upon novel deep learning measurements that allow us not only to trace positive attention through citation but also negative attention through explicit criticism with a novel, comprehensive database of over 20,000 human-validated critical citations. We find that younger scientists tend toward disruptive contributions that push the frontier, while older scientists engage in combinatorial innovation with an aging collection of components. This includes analyzing the impact of the 1994 U.S. Supreme Court ruling on mandatory retirement and examining how unexpected collaborations affect citation patterns.

cs.DL↗

arXiv2025

The (Short-Term) Effects of Large Language Models on Unemployment and Earnings

Large Language Models have spread rapidly since the release of ChatGPT in late 2022, accompanied by claims of major productivity gains but also concerns about job displacement. This paper examines the short-run labor market effects of LLM adoption by comparing earnings and unemployment across occupations with differing levels of exposure to these technologies. Using a Synthetic Difference in Differences approach, we estimate the impact of LLM exposure on earnings and unemployment. Our findings show that workers in highly exposed occupations experienced earnings increases following ChatGPT's introduction, while unemployment rates remained unchanged. These results suggest that initial labor market adjustments to LLMs operate primarily through earnings rather than worker reallocation.

econ.GN↗

arXiv2024

From Division to Unity: A Large-Scale Study on the Emergence of Computational Social Science, 1990-2021

We present a comprehensive study on the emergence of Computational Social Science (CSS) - an interdisciplinary field leveraging computational methods to address social science questions - and its impact on adjacent social sciences. We trained a robust CSS classifier using papers from CSS-focused venues and applied it to 11 million papers spanning 1990 to 2021. Our analysis yielded three key findings. First, there were two critical inflections in the rise of CSS. The first occurred around 2005 when psychology, politics, and sociology began engaging with CSS. The second emerged in approximately 2014 when economics finally joined the trend. Sociology is currently the most engaged with CSS. Second, using the density of yearly knowledge embeddings constructed by advanced transformer models, we observed that CSS initially lacked a cohesive identity. From the early 2000s to 2014, however, it began to form a distinct cluster, creating boundaries between CSS and other social sciences, particularly in politics and sociology. After 2014, these boundaries faded, and CSS increasingly blended with the social sciences. Third, shared data-driven methods homogenized CSS papers across disciplines, with politics and economics showing the most alignment due to the combined influence of CSS and causal identification. Nevertheless, non-CSS papers in sociology, psychology, and politics became more divergent. Taken together, these findings highlight the dynamics of division and unity as new disciplines emerge within existing knowledge landscapes. A live demo of CSS evolution can be found in https://evolution-css.netlify.app/

cs.CY↗

arXiv2024

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 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 China and the U.S., several recent policies have been put in place to discourage AI scientists from migrating to, or collaborating with, the other nation. Nevertheless, the extent of talent migration 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 2000, China and the U.S. have led the field in terms of impact, novelty, productivity, and workforce. Most AI scientists who move to China come from the U.S., and most who move to the U.S. come from China, highlighting a notable bidirectional talent migration. Moreover, the vast majority of those moving in either direction have Asian ancestry. Upon moving, those scientists continue to collaborate frequently with those in the origin country. Although the number of collaborations between the two countries has increased 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 works without the other. These findings suggest that instead of suppressing cross-border migration and collaboration between the two nations, the science could benefit from promoting such activities.

cs.CY↗

arXiv2024

In Silico Sociology: Forecasting COVID-19 Polarization with Large Language Models

By training deep neural networks on massive archives of digitized text, large language models (LLMs) learn the complex linguistic patterns that constitute historic and contemporary discourses. We argue that LLMs can serve as a valuable tool for sociological inquiry by enabling accurate simulation of respondents from specific social and cultural contexts. Applying LLMs in this capacity, we reconstruct the public opinion landscape of 2019 to examine the extent to which the future polarization over COVID-19 was prefigured in existing political discourse. Using an LLM trained on texts published through 2019, we simulate the responses of American liberals and conservatives to a battery of pandemic-related questions. We find that the simulated respondents reproduce observed partisan differences in COVID-19 attitudes in 84% of cases, significantly greater than chance. Prompting the simulated respondents to justify their responses, we find that much of the observed partisan gap corresponds to differing appeals to freedom, safety, and institutional trust. Our findings suggest that the politicization of COVID-19 was largely consistent with the prior ideological landscape, and this unprecedented event served to advance history along its track rather than change the rails.

cs.CY↗

arXiv2024

Learning from One and Only One Shot

Humans can generalize from only a few examples and from little pretraining on similar tasks. Yet, machine learning (ML) typically requires large data to learn or pre-learn to transfer. Motivated by nativism and artificial general intelligence, we directly model human-innate priors in abstract visual tasks such as character and doodle recognition. This yields a white-box model that learns general-appearance similarity by mimicking how humans naturally ``distort'' an object at first sight. Using just nearest-neighbor classification on this cognitively-inspired similarity space, we achieve human-level recognition with only $1$--$10$ examples per class and no pretraining. This differs from few-shot learning that uses massive pretraining. In the tiny-data regime of MNIST, EMNIST, Omniglot, and QuickDraw benchmarks, we outperform both modern neural networks and classical ML. For unsupervised learning, by learning the non-Euclidean, general-appearance similarity space in a $k$-means style, we achieve multifarious visual realizations of abstract concepts by generating human-intuitive archetypes as cluster centroids.

cs.CV↗

arXiv2024

Limited Diffusion of Scientific Knowledge Forecasts Collapse

Market bubbles emerge when asset prices are driven unsustainably higher than asset values and shifts in belief burst them. We demonstrate the same phenomenon for biomedical knowledge when promising research receives inflated attention. We predict deflationary events by developing a diffusion index that captures whether research areas have been amplified within social and scientific bubbles or have diffused and become evaluated more broadly. We illustrate our diffusion approach contrasting the trajectories of cardiac stem cell research and cancer immunotherapy. We then trace the diffusion of unique 28,504 subfields in biomedicine comprising nearly 1.9M papers and more than 80M citations and demonstrate that limited diffusion of biomedical knowledge anticipates abrupt decreases in popularity. Our analysis emphasizes that restricted diffusion, implying a socio-epistemic bubble, leads to dramatic collapses in relevance and attention accorded to scientific knowledge.

cs.SI↗