Benjamin Chin-Yee

Scientists often use generics, that is, unquantified statements about whole categories of people or phenomena, when communicating research findings (e.g., “statins reduce cardiovascular events”). Large language models (LLMs), such as ChatGPT, frequently adopt the same style when summarizing scientific texts. However, generics can prompt overgeneralizations, especially when they are interpreted differently across audiences. In a study comparing laypeople, scientists, and two leading LLMs (ChatGPT-5 and DeepSeek), we found systematic differences in interpretation of generics. Compared to most scientists, laypeople judged scientific generics as more generalizable and credible, while LLMs rated them even higher. These mismatches highlight significant risks for science communication. Scientists may use generics and incorrectly assume laypeople share their interpretation, while LLMs may systematically overgeneralize scientific findings when summarizing research. Our findings underscore the need for greater attention to language choices in both human and LLM-mediated science communication.

Biomarkers are central to the practice of precision oncology, which looks to novel biomarkers to ‘personalize’ cancer care. Philosophers have highlighted epistemic issues surrounding biomarkers but a general account of their role in clinical reasoning is lacking. This article examines biomarker use in clinical reasoning through the lens of abstraction. I propose clinical abstraction as a descriptive and normative account of reasoning with biomarkers that overcomes epistemic and ethical problems raised in the literature.

With increasing detection of early cancers and precancers, debate over the cancer label has intensified. Although attention focuses on relabeling solid tumors to reduce overtreatment, hematologic precursor conditions remain overlooked. Clonal cytopenia of undetermined significance (CCUS) carries increased risk of progression to myelodysplastic neoplasms and acute myeloid leukemia, with high-risk CCUS having a prognosis similar to that of lower-risk myelodysplastic neoplasms. Because these conditions are managed with surveillance or systemic therapy based on risk, diagnostic labels and linguistic framing may influence decision-making. We investigated whether these factors are associated with changes in risk perceptions and treatment decisions in CCUS.

Artificial intelligence chatbots driven by large language models (LLMs) have the potential to increase public science literacy and support scientific research, as they can quickly summarize complex scientific information in accessible terms. However, when summarizing scientific texts, LLMs may omit details that limit the scope of research conclusions, leading to generalizations of results broader than warranted by the original study. We tested 10 prominent LLMs, including ChatGPT-4o, ChatGPT-4.5, DeepSeek, LLaMA 3.3 70B, and Claude 3.7 Sonnet, comparing 4900 LLM-generated summaries to their original scientific texts. Even when explicitly prompted for accuracy, most LLMs produced broader generalizations of scientific results than those in the original texts, with DeepSeek, ChatGPT-4o, and LLaMA 3.3 70B overgeneralizing in 26–73% of cases. In a direct comparison of LLM-generated and human-authored science summaries, LLM summaries were nearly five times more likely to contain broad generalizations (OR = 4.85, 95% CI [3.06, 7.70], p < 0.001). Notably, newer models tended to perform worse in generalization accuracy than earlier ones. Our results indicate a strong bias in many widely used LLMs towards overgeneralizing scientific conclusions, posing a significant risk of large-scale misinterpretations of research findings. We highlight potential mitigation strategies, including lowering LLM temperature settings and benchmarking LLMs for generalization accuracy.

Should we say 'a little bit of cancer'? Many argue that we should avoid the phrase and instead relabel early cancers as a strategy to prevent overtreatment. Against this, I argue that we should not shy away from saying 'a little bit of cancer', and, moreover, that shying away misses a key opportunity to address the problem of overtreatment. Drawing on speech act theory, I examine the diagnosis of cancer as illocutionary speech act and argue for a revisionist strategy which seeks to shift the normative force of the cancer diagnosis qua speech act. I show how this strategy offers distinct advantages vis-à-vis alternative approaches to tackling the problem of overtreatment.

“Actionability” is a key concept in precision oncology. Its precise definition, however, remains contested. This article undertakes a philosophical analysis of “actionability” to aid in conceptual clarification. We map distinct concepts of actionability, arguing that each is best understood as a contextually objective category articulated to mitigate risk of “conceptual slippage.” We defend “interactive pluralism,” acknowledging the need for distinct concepts but also for conceptual interaction in practice. This article thus offers insights for both practitioners and philosophers, clarifying approaches to actionability for scientists and clinicians and serving as a case study to test competing views on scientific pluralism.

Naturalistic theories of disease appeal to concepts of biological function, and use the notion of dysfunction as the basis of their definitions. Debates in the philosophy of biology demonstrate how attributing functions in organisms and establishing the function-dysfunction distinction is by no means straightforward. This problematization of functional ascription has undermined naturalistic theories and led some authors to abandon the concept of dysfunction, favoring instead definitions based in normative criteria or phenomenological approaches. Although this work has enhanced our understanding of disease and illness, we need not necessarily abandon naturalistic concepts of function and dysfunction in the disease debate. This article attempts to move towards a new naturalistic theory of disease that overcomes the limitations of previous definitions and offers advantages in the clinical setting. Our approach involves a re-evaluation of concepts of biological function employed by naturalistic theories. Drawing on recent insights from the philosophy of biology, we develop a contextual and evaluative account of function that is better suited to clinical medicine and remains consistent with contemporary naturalism. We also show how an updated naturalistic view shares important affinities with normativist and phenomenological positions, suggesting a possibility for consilience in the disease debate.

Naturalistic theories of disease appeal to concepts of biological function, and use the notion of dysfunction as the basis of their definitions. Debates in the philosophy of biology demonstrate how attributing functions in organisms and establishing the function-dysfunction distinction is by no means straightforward. This problematization of functional ascription has undermined naturalistic theories and led some authors to abandon the concept of dysfunction, favoring instead definitions based in normative criteria or phenomenological approaches. Although this work has enhanced our understanding of disease and illness, we need not necessarily abandon naturalistic concepts of function and dysfunction in the disease debate. This article attempts to move towards a new naturalistic theory of disease that overcomes the limitations of previous definitions and offers advantages in the clinical setting. Our approach involves a reevaluation of concepts of biological function employed by naturalistic theories. Drawing on recent insights from the philosophy of biology, we develop a contextual and evaluative account of function that is better suited to clinical medicine and remains consistent with contemporary naturalism. We also show how an updated naturalistic view shares important affinities with normativist and phenomenological positions, suggesting a possibility for consilience in the disease debate.

The COVID-19 global pandemic has stretched the capacities of public health institutions and health systems around the world, opening the door to a range of technologically-driven solutions. In this article, we seek to historicize the expanding role of digital health technologies and examine the political-economic context from which they have emerged. Drawing on critical insights from science and technology studies, we maintain that the rise of digital health technologies has been catalyzed by broad shifts in global health governance that have expanded the role of market forces in public health and a unique set of political and economic crises that have accelerated the adoption of digital technologies—often under the guise of appeals to technological innovation to address “unprecedented” crises. These interrelated historical trends, we contend, are critical for understanding current state responses to the pandemic and possibilities for more equitable and democratic applications of technology in public health.

ABSTRACT: Generalizations in medical research can be informative, but also misleading. Building on recent work in the philosophy of science and ethics of communication, I offer a novel analysis of common generalizations in clinical trials as generics in natural language. Generics, which express generalizations without terms of quantification, have attracted considerable attention from philosophers, psychologists, and linguists. My analysis draws on probabilistic and contextual features of generics to diagnose how these generalizations function and malfunction across communicative contexts in medicine. Given a high risk of misinterpretation ("slippage"), I recommend avoidance of generic claims about medical interventions in public contexts, exemplified by clinical trials and medical research more generally. Generics should only be used with vigilance in private contexts, exemplified by the physician–patient encounter. My analysis provides tools to support vigilance when communicating with generics, suggests new norms for public science communication, and raises deeper questions in the ethics of clinical communication.

Cancers rely on multiple, heterogeneous processes at different scales, pertaining to many biomedical fields. Therefore, understanding cancer is necessarily an interdisciplinary task that requires placing specialised experimental and clinical research into a broader conceptual, theoretical, and methodological framework. Without such a framework, oncology will collect piecemeal results, with scant dialogue between the different scientific communities studying cancer. We argue that one important way forward in service of a more successful dialogue is through greater integration of applied sciences (experimental and clinical) with conceptual and theoretical approaches, informed by philosophical methods. By way of illustration, we explore six central themes: (i) the role of mutations in cancer; (ii) the clonal evolution of cancer cells; (iii) the relationship between cancer and multicellularity; (iv) the tumour microenvironment; (v) the immune system; and (vi) stem cells. In each case, we examine open questions in the scientific literature through a philosophical methodology and show the benefit of such a synergy for the scientific and medical understanding of cancer.

Minimal residual disease (MRD), a measure of residual cancer cells, is a concept increasingly employed in precision oncology, touted as a key predictive biomarker to guide treatment decisions. This paper critically analyzes the expanding role of MRD as a predictive biomarker in hematologic cancers. I outline the argument for MRD as a predictive biomarker, articulating its premises and the empirical conditions that must hold for them to be true. I show how these conditions, while met in paradigmatic cases of MRD use in cancer, may not hold across other cancers where MRD is currently being applied, weakening the argument that MRD serves as an effective predictive biomarker across cancer medicine.

It is unknown to what extent medical researchers generalize study findings beyond their samples when their sample size, sample diversity, or knowledge of conditions that support external validity do not warrant it. It is also unknown to what extent medical researchers describe their results with precise quantifications or unquantified generalizations, i.e., generics, that can obscure variations between individuals. We therefore systematically reviewed all prospective studies (n = 533) published in the top four highest ranking medical journals, Lancet, New England Journal of Medicine (NEJM), Journal of the American Medical Association (JAMA), and the British Medical Journal (BMJ), from January 2022 to May 2023. We additionally reviewed all NEJM Journal Watch clinical research summaries (n = 143) published during the same time. Of all research articles reporting prospective studies, 52.5% included generalizations beyond specific national study populations, with the numbers of articles with generics varying significantly between journals (JAMA = 12%; Lancet = 77%) (p < 0.001, V = 0.48). There was no evidence that articles containing broader generalizations or generics were correlated with larger or more nationally diverse samples. Moreover, only 10.2% of articles with generalizations beyond specific national populations reported external validity strengthening factors that could potentially support such extrapolations. There was no evidence that original research articles and NEJM Journal Watch summaries intended for practitioners differed in their use of broad generalizations, including generics. Finally, from the journal with the highest citation impact, articles containing broader conclusions were correlated with more citations. Since there was no evidence that studies with generalizations beyond specific national study populations or with generics were associated with larger, more nationally diverse samples, or with reports of population similarity that may permit extensions of conclusions, our findings suggest that the generalizations in many articles were insufficiently supported. Caution against overly broad generalizations in medical research is warranted.

We live in the Age of Big Data. In medicine, artificial intelligence and machine learning algorithms, fueled by big data, promise to change how physicians make diagnoses, determine prognoses, and develop new treatments. An exponential rise in articles on these topics is seen in the medical literature. Recent applications range from the use of deep learning neural networks to diagnose diabetic retinopathy and skin cancer from image databases, to the use of various machine learning algorithms for prognostication in cancer and cardiovascular disease. Many factors are driving the adoption of AI in health care, from...