Andrés Páez

El riesgo epistémico en el derecho es la posibilidad de cometer errores o aceptar conclusiones falsas debido a la incertidumbre fáctica y a los valores no epistémicos que rodean la toma de decisiones. Uno de los factores que contribuye a la distorsión de las decisiones en todos los momentos procesales es la influencia innegable de los sesgos cognitivos. En este artículo explico cómo estos contribuyen a aumentar el riesgo epistémico y exploro las posibles razones por las que los sistemas judiciales no han tomado medidas significativas para eliminar esta fuente de error. Al final analizo el problema de cómo atribuir responsabilidades a los diferentes agentes que integran el sistema judicial ante una actitud que solo puede ser catalogada como negligencia epistémica.

The opaqueness of many complex machine learning algorithms is often mentioned as one of the main obstacles to the ethical development of artificial intelligence (AI). But what does it mean for an algorithm to be opaque? Highly complex algorithms such as artificial neural networks process enormous volumes of data in parallel along multiple hidden layers of interconnected nodes, rendering their inner workings epistemically inaccessible to any human being, including their designers and developers; they are ‘black boxes’ for all their stakeholders. But opaqueness is not always the inevitable result of technical complexity. Sometimes, the way an algorithm works is intentionally hidden from view for proprietary reasons, especially in commercial automated decision systems, creating an entirely different type of opaqueness. In the first part of the chapter, we will examine these two ways of understanding opacity and the ethical implications that stem from each of them. In the second part, we explore the different explanatory methods that have been developed in computer science to overcome an AI system’s technical opaqueness. As the analysis shows, explainable AI (XAI) still faces numerous challenges

El concepto de reputación ocupa un lugar central en muchas esferas de la vida humana, incluyendo el derecho, pero no existen muchos análisis filosóficos del mismo. En este ensayo presento un análisis de la reputación desde la perspectiva de la epistemología social, en particular, haciendo uso de las teorías contemporáneas sobre actitudes epistémicas grupales. El ensayo comienza examinando las interpretaciones del concepto de reputación que se encuentran en la literatura jurídica. En particular, examino el bien conocido análisis del concepto que hace Post (1986) en el contexto de la ley de difamación. A partir de la tercera sección presento el análisis epistemológico y establezco una distinción entre reputación objetiva y subjetiva, una distinción que es fácilmente identificable en la práctica. Mi interés principal reside en el concepto de reputación subjetiva, que es el que presenta el mayor desafío conceptual desde el punto de vista epistemológico. En la sección 4 emprendo la tarea de determinar si podemos entender la reputación subjetiva como un tipo de creencia grupal. Tras rechazar esta posibilidad, en la sección final ofrezco una manera diferente de entender este concepto y examino sus implicaciones para el derecho.

Uno de los principios éticos mencionados más frecuentemente en los lineamientos para el desarrollo de la inteligencia artificial (IA) es la transparencia algorítmica. Sin embargo, no existe una definición estándar de qué es un algoritmo transparente ni tampoco es evidente por qué la opacidad algorítmica representa un reto para el desarrollo ético de la IA. También se afirma a menudo que la transparencia algorítmica fomenta la confianza en la IA, pero esta aseveración es más una suposición a priori que una tesis basada en evidencia empírica. Tampoco se discute mucho hasta qué punto es técnicamente posible volver transparente la caja negra de la IA a través de los métodos de explicabilidad. En este capítulo haré un análisis de la interrelación entre los conceptos de transparencia, explicabilidad y confianza. Inicialmente, analizaré los diferentes tipos de opacidad algorítmica para entender mejor cuál es el problema al que nos enfrentamos. En las secciones subsiguientes trazaré la relación entre explicabilidad y transparencia, y presentaré los límites de los métodos actuales de explicabilidad. En la sección final, examinaré la evidencia empírica acerca de la relación entre transparencia y confianza en las decisiones automatizadas basadas en sistemas de IA.

In the natural and social sciences, it is common to use toy models—extremely simple and highly idealized representations—to understand complex phenomena. Some of the simple surrogate models used to understand opaque machine learning (ML) models, such as rule lists and sparse decision trees, bear some resemblance to scientific toy models. They allow non-experts to understand how an opaque ML model works globally via a much simpler model that highlights the most relevant features of the input space and their effect on the output. The obvious difference is that the common target of a toy and a full-scale model in the sciences is some phenomenon in the world, while the target of a surrogate model is another model. This essential difference makes toy surrogate models (TSMs) a new object of study for theories of understanding, one that is not easily accommodated under current analyses. This paper provides an account of what it means to understand an opaque ML model globally with the aid of such simple models.

To verify the occurrence of a singular instance of testimonial injustice three facts must be established. The first is whether the hearer in fact has an identity prejudice of which she may or may not be aware; the second is whether that prejudice was in fact the cause of the unjustified credibility deficit; and the third is whether there was in fact a credibility deficit in the testimonial exchange. These three elements constitute the facts of the matter of testimonial injustice. In this essay we argue that none of these facts can be established with any degree of confidence, and therefore that testimonial injustice is an undetectable phenomenon in singular instances. Our intention is not to undermine the idea of testimonial injustice, but rather to set limits to what can be justifiably asserted about it. According to our argument, although there are insufficient reasons to identify individual acts of testimonial injustice, it is possible to recognize recurrent patterns of epistemic responses to speakers who belong to specific social groups. General testimonial injustice can thus be characterized as a behavioral tendency of a prejudiced hearer.

La literatura sobre razonamiento probatorio busca incidir en la determinación de los hechos en los procesos judiciales. Para alcanzar dicho propósito, no basta con dirigir la mirada hacia disciplinas extrajurídicas exitosas e integrar lo que de ellas pueda extraerse a las teorías jurídicas de la prueba y a la práctica judicial. Es necesario, además, considerar el tipo de hechos a probar, los roles de las reglas jurídicas aplicables, y asumir que litigantes y jueces, actuando en un contexto institucional, podrán ser más o menos agudos en la discusión fáctica, entre otras cosas. Minimizar la influencia de tales factores expone a los estudios probatorios a tener un bajo rendimiento práctico. Algo de esto ha ocurrido con la llamada concepción racionalista de la prueba. Su innegable aporte al debate académico en español no ha ido a la par de su impacto judicial. Para explicar esta brecha entre teoría y práctica, identificamos tres creencias-obstáculo que afectan a su discurso sobre la prueba y que deben ser sorteadas si lo que se busca es influir en la forma en que se juzgan los hechos en un sistema judicial realmente existente.

In a recent paper, Erasmus et al. (2021) defend the idea that the ambiguity of the term “explanation” in explainable AI (XAI) can be solved by adopting any of four different extant accounts of explanation in the philosophy of science: the Deductive Nomological, Inductive Statistical, Causal Mechanical, and New Mechanist models. In this chapter, I show that the authors’ claim that these accounts can be applied to deep neural networks as they would to any natural phenomenon is mistaken. I also provide a more general argument as to why the notion of explainability as it is currently used in the XAI literature bears little resemblance to the traditional concept of scientific explanation. It would be more fruitful to use the label “understandable AI” to avoid the confusion that surrounds the goal and purposes of XAI. In the second half of the chapter, I argue for a pragmatic conception of understanding that is better suited to play the central role attributed to explanation in XAI. Following De Regt (2017) and Kuorikoski and Ylikoski (2015), the conditions of satisfaction for understanding an ML system are fleshed out in terms of an agent’s success in using the system.

Cada día es más evidente que existen muchas formas sutiles de exclusión y parcialidad que afectan el correcto funcionamiento de los sistemas jurídicos. El concepto de injusticia epistémica, introducido por la filósofa Miranda Fricker, ofrece una herramienta conceptual útil para comprender estas formas de exclusión y parcialidad judicial que a menudo pasan desapercibidas. En este artículo presentamos la teoría original de Fricker y algunas de las aplicaciones del concepto de injusticia epistémica en los procesos jurídicos. En particular, queremos demostrar que la semilla sembrada por Fricker ha generado un floreciente campo de estudio en el cual el concepto se utiliza para analizar muchos fenómenos diferentes en el ámbito jurídico, no siempre siguiendo la caracterización original proporcionada por ella. Esto ha llevado a una distinción entre lo que llamaremos la versión estrecha del concepto, que se acerca a la descripción original de Fricker, y la versión más amplia de injusticia epistémica, que es una noción más controvertida porque existe el riesgo de que se confunda con otros conceptos más conocidos como el sexismo, la discriminación racial, la opresión, el silenciamiento y el gaslighting. Mostraremos que el valor de la versión estrecha es principalmente teórico y que para utilizar el concepto de injusticia epistémica se debe adoptar una comprensión más liberal del mismo.

As systems based on opaque Artificial Intelligence (AI) continue to flourish in diverse real-world applications, understanding these black box models has become paramount. In response, Explainable AI (XAI) has emerged as a field of research with practical and ethical benefits across various domains. This paper not only highlights the advancements in XAI and its application in real-world scenarios but also addresses the ongoing challenges within XAI, emphasizing the need for broader perspectives and collaborative efforts. We bring together experts from diverse fields to identify open problems, striving to synchronize research agendas and accelerate XAI in practical applications. By fostering collaborative discussion and interdisciplinary cooperation, we aim to propel XAI forward, contributing to its continued success. Our goal is to put forward a comprehensive proposal for advancing XAI. To achieve this goal, we present a manifesto of 27 open problems categorized into nine categories. These challenges encapsulate the complexities and nuances of XAI and offer a road map for future research. For each problem, we provide promising research directions in the hope of harnessing the collective intelligence of interested stakeholders.

Consensus in the contemporary philosophical literature has it that conserved quantity theories of causation such as that of Dowe [2000]—according to which causation is to be analysed in terms of the exchange of conserved quantities (e.g., energy)—face damning problems when confronted with contemporary physics, where the notion of conservation becomes delicate. In particular, in general relativity it is often claimed that there simply are no conservation laws for (say) total-stress energy. If this claim is correct, it is difficult to see how conserved quantity theories of causation could survive. In this article, we resist the above consensus and defend conserved quantity theories from this conclusion, at least when focusing on the apparent problems posed by general relativity. We argue that this approach to causation can continue to be defended in general relativity, once one appreciates (a) the availability of approximate symmetries in generic general relativistic spacetimes, and (b) the role of modelling and idealisation in that theory. Given these points, conserved quantity theories of causation must stand or fall on other grounds.

Resumen: Una de las formas más comunes y menos estudiadas de parcialidad judicial subjetiva es la disminución de la credibilidad otorgada a un testigo debido a un prejuicio identitario implícito del agente judicial. En la epistemología social, este fenómeno ha sido estudiado bajo la rúbrica de la injusticia testimonial. En este ensayo mostramos que para determinar la ocurrencia de un caso de injusticia testimonial en el derecho se deben cumplir tres condiciones que son imposibles de verificar empíricamente y que están basadas en presupuestos psicológicos que han sido puestos en duda en años recientes. Sin la posibilidad de verificar estas condiciones, estamos frente a un tipo de parcialidad judicial indetectable. En su lugar, ofrecemos una nueva forma de entender la injusticia testimonial en el derecho como un fenómeno más general que describe una tendencia comportamental prejuiciosa recurrente por parte de un agente judicial. Entendida de este modo, la injusticia testimonial es completamente verificable. Abstract: One of the most common and least studied forms of subjective judicial partiality is the credibility deficit suffered by a witness due to an implicit identity prejudice in a judicial agent. Social epistemologists have studied this form of partiality under the rubric, “testimonial injustice.” In this essay we argue that to verify the occurrence of a singular instance of testimonial injustice in law three facts must be established, none of which can be established with any degree of confidence. Furthermore, all of them are based on psychological suppositions that have been recently discredited. Without the possibility of establishing these facts, we end up with an undetectable form of subjective judicial partiality. In its place, we offer a new way of understanding testimonial injustice in law as a general phenomenon that describes a recurrent pattern of prejudiced behavior in a judicial agent. Under this characterization, testimonial injustice becomes empirically verifiable.

There is a growing awareness that there are many subtle forms of exclusion and partiality that affect the correct workings of a judicial system. The concept of epistemic injustice, introduced by the philosopher Miranda Fricker, is a useful conceptual tool to understand forms of judicial partiality that often go undetected. In this paper, we present Fricker’s original theory and some of the applications of the concept of epistemic injustice in legal processes. In particular, we want to show that the seed planted by Fricker has flourished into a rich field of study in which the concept is used to analyze many different phenomena in law, not always following the original characterization provided by her. This has led to a distinction between what we will call the narrow version of the concept, which is closer to Fricker’s original account, and the wider version of epistemic injustice, which is a more controversial notion because it is always on the verge of morphing into other well-known concepts like sexism, racial discrimination, oppression, silencing, and gaslighting. We will show that the value of the narrow version is mostly theoretical, and that in order to use the concept of epistemic injustice one must adopt a more liberal understanding of it.

Rules 405(a) and 608(a) of the Federal Rules of Evidence allow the use of testimony about a witness’s reputation to support or undermine his or her credibility in trial. This paper analyzes the evidential weight of such testimony from the point of view of social epistemology and the theory of social networks. Together they provide the necessary elements to analyze how reputation is understood in this case, and to assess the epistemic foundation of a reputational attribution. The result of the analysis will be that reputational testimony is extremely weak from an epistemological point of view, and that in many cases there are more reliable substitutes that achieve a similar purpose. The obvious fix, in my view, is to eliminate the use of reputation testimony to support or undermine the credibility, honesty, chastity or peacefulness of a witness.

The use of machine learning algorithms has become ubiquitous in hiring decisions. Recent studies have shown that many of these algorithms generate unlawful discriminatory effects in every step of the process. The training phase of the machine learning models used in these decisions has been identified as the main source of bias. For a long time, discrimination cases have been analyzed under the banner of disparate treatment and disparate impact, but these concepts have been shown to be ineffective in the context of AI. This paper examines the possibility of studying algorithmic discrimination from the perspective of negligence law. Negligent selection and validation of datasets and classifiers is a salient cause of improper training in machine learning. It is a breach of an employer’s duty to protect others, a failure of its exercise of due care in the manner of choosing employees. Adopting negligence to algorithmic discrimination requires a reformulation of what is foreseeable and reasonable within the context of black box algorithms, and a shift in the burden of proof to the employer, who must present evidence of its efforts to avoid bias.

El uso de sistemas de machine learning en los procesos de selección laboral ha sido de gran utilidad para agilizarlos y volverlos más eficientes, pero al mismo tiempo ha generado problemas en términos de equidad, confiabilidad y transparencia. En este artículo comenzamos explicando los diferentes usos de la Inteligencia Artificial en los procesos de selección laboral en Estados Unidos. Presentamos los sesgos sexuales y raciales que han sido detectados en algunos de ellos y explicamos los obstáculos jurídicos y prácticos para su detección y análisis. Un posible remedio jurídico a la discriminación algorítmica es el régimen de impacto diferencial sistémico utilizado en Estados Unidos. Usamos algunas características de este régimen para señalar los vacíos jurídicos frente a la discriminación en el acceso al empleo en el ámbito del derecho laboral colombiano, y ofrecemos algunas características generales que podría tener un régimen de esa naturaleza.

This paper raises three questions regarding the attribution of beliefs, desires, and intentions to robots. The first one is whether humans in fact engage in robot mindreading. If they do, this raises a second question: does robot mindreading foster trust towards robots? Both of these questions are empirical, and I show that the available evidence is insufficient to answer them. Now, if we assume that the answer to both questions is affirmative, a third and more important question arises: should developers and engineers promote robot mindreading in view of their stated goal of enhancing transparency? My worry here is that by attempting to make robots more mind-readable, they are abandoning the project of understanding automatic decision processes. Features that enhance mind-readability are prone to make the factors that determine automatic decisions even more opaque than they already are. And current strategies to eliminate opacity do not enhance mind-readability. The last part of the paper discusses different ways to analyze this apparent trade-off and suggests that a possible solution must adopt tolerable degrees of opacity that depend on pragmatic factors connected to the level of trust required for the intended uses of the robot.

In this paper I argue that the search for explainable models and interpretable decisions in AI must be reformulated in terms of the broader project of offering a pragmatic and naturalistic account of understanding in AI. Intuitively, the purpose of providing an explanation of a model or a decision is to make it understandable to its stakeholders. But without a previous grasp of what it means to say that an agent understands a model or a decision, the explanatory strategies will lack a well-defined goal. Aside from providing a clearer objective for XAI, focusing on understanding also allows us to relax the factivity condition on explanation, which is impossible to fulfill in many machine learning models, and to focus instead on the pragmatic conditions that determine the best fit between a model and the methods and devices deployed to understand it. After an examination of the different types of understanding discussed in the philosophical and psychological literature, I conclude that interpretative or approximation models not only provide the best way to achieve the objectual understanding of a machine learning model, but are also a necessary condition to achieve post hoc interpretability. This conclusion is partly based on the shortcomings of the purely functionalist approach to post hoc interpretability that seems to be predominant in most recent literature.

Moore’s Paradox is a test case for any formal theory of belief. In Knowledge and Belief, Hintikka developed a multimodal logic for statements that express sentences containing the epistemic notions of knowledge and belief. His account purports to offer an explanation of the paradox. In this paper I argue that Hintikka’s interpretation of one of the doxastic operators is philosophically problematic and leads to an unnecessarily strong logical system. I offer a weaker alternative that captures in a more accurate way our logical intuitions about the notion of belief without sacrificing the possibility of providing an explanation for problematic cases such as Moore’s Paradox.

Esta breve introducción a la filosofía de la ciencia parte del hecho de que tanto la investigación científica como el razonamiento probatorio judicial tienen un carácter inductivo. En esa medida, comparten características esenciales que permiten que el derecho se nutra de muchas de las reflexiones de la filosofía de la ciencia. El capítulo se concentra en cuatro temas principales: los criterios de demarcación entre el conocimiento científico y la pseudociencia; el carácter derrotable de las conclusiones de la ciencia y el derecho; la estructura corroborativa de la relación entre la evidencia y las hipótesis científicas y judiciales; y las teorías de la explicación tanto en el ámbito científico como en el judicial.

Los sesgos cognitivos afectan negativamente la toma de decisiones en todas las esferas de la vida, incluyendo las decisiones de los jueces. La imposibilidad de eliminarlos por completo de la práctica del derecho, o incluso de controlar sus efectos, contrasta con el anhelo de que las decisiones judiciales sean el resultado exclusivo de un razonamiento lógico-jurídico correcto. Frente el efecto sistemático, recalcitrante y porfiado de los sesgos cognitivos, una posible estrategia para disminuir su efecto es enfocarse, no en modificar el comportamiento de los agentes judiciales, sino en moldear su entorno para que los sesgos tengan menos oportunidades de aparecer. En particular, es más provechoso concentrar los esfuerzos en identificar aquellas reglas procesales que faciliten o limiten el efecto de los sesgos. En este ensayo propongo un diálogo fructífero entre el derecho y la psicología que puede llevar a mejorar la calidad de los sistemas judiciales. La propuesta está basada en la creación de guías de práctica judicial que puedan servir de base para la creación de protocolos procesales que limiten en la medida de lo razonable sus efectos negativos y conduzcan a mejores decisiones judiciales.

Testimony about the future dangerousness of a person has become a central staple of many judicial processes. In settings such as bail, sentencing, and parole decisions, in rulings about the civil confinement of the mentally ill, and in custody decisions in a context of domestic violence, the assessment of a person’s propensity towards physical or sexual violence is regarded as a deciding factor. These assessments can be based on two forms of expert testimony: actuarial or clinical. The purpose of this paper is to examine the scientific and epistemological basis of both methods of prediction or risk assessment. My analysis will reveal that this kind of expert testimony is scientifically baseless. The problems I will discuss will generate a dilemma for factfinders: on the one hand, given the weak predictive abilities of the branches of science involved, they should not admit expert clinical or actuarial testimony as evidence; on the other hand, there is a very strong tradition and a vast jurisprudence that supports the continued use of this kind of expert testimony. It is a clear case of the not so uncommon conflict between science and legal tradition.

There are several approaches to the way in which standards of proof in law should be understood. In the central article in this issue, Rodrigo Coloma suggests that they should be conceived either as thresholds either as prototypes. He also offers a taxonomy of the different uses of standards of proof, which transcends their function as error distributors. In their comments on the paper, Raymundo Gama and Claudio Agüero clarify and challenge this way of understanding standards of proof. Some of their criticisms are right on target, but many others are directed more at Coloma’s theoretical presuppositions than at his substantial contributions. In this paper I offer a wider context to evaluate their disagreements.

In many contexts, such as business, science and law, it is essential to determine whether a company, a product or a person in fact has the reputation attributed to it, regardless of whether that reputation has been rightly earned. In this paper I offer two necessary and jointly sufficient conditions for the attribution of a reputation to a subject within a social group. The first one concerns the way in which reputational information is spread among the members of the relevant social group; the second is the existence, in most members of the group, of a disposition to interpret and evaluate a reputational object in a manner molded by reputational information. Reputational claims generate expectations that are tested every time a member of the group comes in contact with the reputational object. In many cases, the preferential search for behavioral evidence consistent with those expectations will result in a distorted evaluation of the reputational object.

El desarrollo de las artes visuales durante el siglo XX desdibujó la frontera entre aquellos objetos y artefactos que llamamos obras de arte, y aquéllos que no son merecedores de ese título. Arthur Danto ha propuesto una teoría estética a la luz de la cual sería posible volver a definir los límites del arte. En este ensayo examino dos de los aspectos más problemáticos de la teoría: la importancia excesiva que Danto le otorga al concepto de mímesis y su concepción teleológica de la historia del arte. Si atenuamos el papel de estos dos elementos, la teoría pierde gran parte de su poder explicativo. En las últimas dos secciones del ensayo arguyo que el mundo del arte al que Danto atribuye un papel central en el problema de la demarcación debe él mismo ser considerado un subproducto de las condiciones que determinan la naturaleza del arte en general.

Durante los últimos años, el problema de cómo justificar aquellas creencias que se originan en testimonios ha ocupado un lugar central en la epistemología. Sin embargo, muy pocas de esas reflexiones son conocidas en el derecho probatorio. En el presente ensayo analizo la prueba testimonial a la luz de estas reflexiones con el fin de poner de manifiesto los supuestos epistemológicos de algunos principios procesales. En concreto, analizo la legislación colombiana y la estadounidense en el marco de la disputa entre el reduccionismo y el antirreduccionismo en la filosofía del testimonio. Al final sugiero una nueva aproximación epistemológica que replantea los términos de la disputa y que tiene implicaciones importantes para la valoración de la prueba testimonial en los procesos judiciales.