Lorenzo Sartori

In a recent contribution, Boisseau (2026) argues against a popular analogy between artificial intelligence (AI) and human experts. We start by clarifying that the analogy does not aim to establish that AI is an expert, but rather that AI can be a potential source of epistemic authority and trust. Once this is clarified, we examine Boisseau’s objections against the AI-expert analogy and demonstrate that they are unsuccessful. Finally, we contend that AI should be seen as an interestingly novel case where the categories “scientific instrument” and “expert” are not necessarily mutually exclusive.

In their investigation on the epistemic role of scientific models, many philosophers of science have historically focused on scientific representation. In this received view, a model serves as a system that represents a target system – a portion of phenomena of interest. However, this representational view has been criticised, particularly by proponents of so-called artefactualism. This view contrasts models’ artefactual nature with the representational perspective. This paper has two aims. First, it clarifies the points of disagreement between artefactualism and the representational view I support, which bases representation on interpretation, denotation, exemplification and de-idealisation functions – rather than similarity or isomorphism. Second, it offers a synergistic solution that incorporates several insights of artefactualism within a more sophisticated representational account.

In this paper, I argue that model organisms (MOs) function as representations of other organisms, in the same way in which scientific models function as representations of their targets. This offers a response to the question of how we justify inferences from MOs to other biological systems. Building on Ankeny and Leonelli's (2020) account of MOs and drawing on the resources of the DEKI account of scientific representation (Frigg and Nguyen 2020), I argue that MO-based inferences are justified iff they exemplify properties that are translated into the ones imputed to the target system by an appropriate mapping function. Then, I defend this account against the charges of Levy and Currie (2015) and Weber (2004), who have proposed non-representational accounts of MOs.

Philosophers have debated at length the epistemological status of scientific thought experiments. I contend that the literature on this topic still lacks a common conceptual framework, a lacuna that produces radical disagreement among the participants in this debate. To remedy this problem, I suggest focusing on the distinction between the internal and the external validity of an experiment, which is also crucial for thought experiments. I then develop an account of both kinds of validity in the context of thought experiments. I show that we can naturally conceptualise internal validity in terms of games of make-believe. Then, I argue that external validity is best defined as accurate representation of a target system. Finally, I turn back to the current debate on thought experiments and show that my diarchic account provides a general framework that can be shared by the competing philosophical views, as well as a fruitful guide for their reconciliation.

Scientists often make use of epistemic representations in order to perform investigations about the real world. So far, philosophers of science interested in epistemic representation of this sort have mostly focused on scientific models. In this thesis, I argue that there are other interesting instances of representation besides models: thought experiments, experimental organisms, and mechanically-produced pictures. These represent portions of the world in the same way as models do, if the concept of epistemic representation is properly understood. In Chapter 1, I introduce and develop the idea that a system functions as an epistemic representation of a designated target system insofar as it is interpreted as a symbol that both possesses and highlights some properties that are then to be imputed to that target system via a proper de-idealising function. In the main body of the thesis, I analyse the above-mentioned types of representations and argue that they function as representations in the same way as scientific models. Chapter 2 discusses the use of thought experiments in physics, with particular focus on Galileo’s thought experiment that illustrates the principle of inertia. Chapter 3 focuses on experimental organism research in biology, with specific attention to the Drosophila melanogaster. Finally, Chapter 4 is about the epistemic use of mechanically produced pictures with the picture of a black hole as the primary case study. In each chapter, I further show that by studying these types of representation through the lens of the account developed in Chapter 1, one can (dis)solve specific issues about thought experiments, model organisms, and pictures respectively. In Chapter 5, I further defend my proposal from scepticism about the concept of representation and its application to these different cases. Chapter 6 concludes the thesis by illustrating the most general implications of my analysis and proposing routes for future enquiry.

In this paper, I first argue that similarity accounts of scientific pictures fail with more realistic cases of scientific pictures. My primary case study is the picture of a black hole, to which I apply an interpretation-based account of picture representation analogous to how models represent: a picture represents a designated target system iff, once interpreted, it exemplifies properties that are imputed to the target via a de-idealising function. Then, I argue that the justification of the inferences from mechanically produced pictures depends on their causal mechanisms of production, in contrast with the standard justificatory strategies employed for model inferences.