Sylvia Wenmackers

Andres Algaba, Carmen Mazijn, Vincent Holst et al.

Citation practices are crucial in shaping the structure of scientific knowledge, yet they are often influenced by contemporary norms and biases. The emergence of Large Language Models (LLMs) introduces a new dynamic to these practices. Interestingly, the characteristics and potential biases of references recommended by LLMs that entirely rely on their parametric knowledge, and not on search or retrieval-augmented generation, remain unexplored. Here, we analyze these characteristics in an experiment using a dataset from AAAI, NeurIPS, ICML, and ICLR, published after GPT-4's knowledge cut-off date. In our experiment, LLMs are tasked with suggesting scholarly references for the anonymized in-text citations within these papers. Our findings reveal a remarkable similarity between human and LLM citation patterns, but with a more pronounced high citation bias, which persists even after controlling for publication year, title length, number of authors, and venue. The results hold for both GPT-4, and the more capable models GPT-4o and Claude 3.5 where the papers are part of the training data. Additionally, we observe a large consistency between the characteristics of LLM's existing and non-existent generated references, indicating the model's internalization of citation patterns. By analyzing citation graphs, we show that the references recommended are embedded in the relevant citation context, suggesting an even deeper conceptual internalization of the citation networks. While LLMs can aid in citation generation, they may also amplify existing biases, such as the Matthew effect, and introduce new ones, potentially skewing scientific knowledge dissemination.

Andres Algaba, Vincent Holst, Floriano Tori et al.

The spread of scientific knowledge depends on how researchers discover and cite previous work. The adoption of large language models (LLMs) in the scientific research process introduces a new layer to these citation practices. However, it remains unclear to what extent LLMs align with human citation practices, how they perform across domains, and may influence citation dynamics. Here, we show that LLMs systematically reinforce the Matthew effect in citations by consistently favoring highly cited papers when generating references. This pattern persists across scientific domains despite significant field-specific variations in existence rates, which refer to the proportion of generated references that match existing records in external bibliometric databases. Analyzing 274,951 references generated by GPT-4o for 10,000 papers, we find that LLM recommendations diverge from traditional citation patterns by preferring more recent references with shorter titles and fewer authors. Emphasizing their content-level relevance, the generated references are semantically aligned with the content of each paper at levels comparable to the ground truth references and display similar network effects while reducing author self-citations. These findings illustrate how LLMs may reshape citation practices and influence the trajectory of scientific discovery by reflecting and amplifying established trends. As LLMs become more integrated into the scientific research process, it is important to understand their role in shaping how scientific communities discover and build upon prior work.

Sylvia Wenmackers

Philosophers of technology have recently started paying more attention to the environmental impacts of AI, in particular of large language models (LLMs) and generative AI (genAI) applications. Meanwhile, few developers of AI give concrete estimates of the ecological impact of their models and products, and even when they do so, their analysis is often limited to green house gas emissions of certain stages of AI development or use. The current proposal encourages practically viable analyses of the sustainability aspects of genAI informed by philosophical ideas.

Leander Vignero, Sylvia Wenmackers

In this paper, we take a fresh look at three Popperian concepts: riskiness, falsifiability, and truthlikeness (or verisimilitude) of scientific hypotheses or theories. First, we make explicit the dimensions that underlie the notion of riskiness. Secondly, we examine if and how degrees of falsifiability can be defined, and how they are related to various dimensions of the concept of riskiness as well as the experimental context. Thirdly, we consider the relation of riskiness to (expected degrees of) truthlikeness. Throughout, we pay special attention to probabilistic theories and we offer a tentative, quantitative account of verisimilitude for probabilistic theories.