Automating discovery: what can we learn from the study of abductive reasoning?

The aim of this paper is to investigate the extent to which abductive inference can be automated. In order to do so, we present the Peircean account of abduction, according to which abduction is the process of generating and selecting an explanatory hypothesis that guides future inquiry (CP 5.171; 1...

Descripción completa

Detalles Bibliográficos
Autores: Rodrigues, Mariana Vitti, Gonzalez, Maria Eunice Quilici
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2024
País:Brasil
Institución:Pontifícia Universidade Católica de São Paulo (PUC-SP)
Repositorio:Cognitio (São Paulo. Online)
Idioma:inglés
OAI Identifier:oai:ojs.pkp.sfu.ca:article/68263
Acceso en línea:https://revistas.pucsp.br/index.php/cognitiofilosofia/article/view/68263
Access Level:acceso abierto
Palabra clave:Abdução
AI-Descartes
Automação
Inferência à melhor explicação
Abduction
Automation
Inference to the best explanation
Descripción
Sumario:The aim of this paper is to investigate the extent to which abductive inference can be automated. In order to do so, we present the Peircean account of abduction, according to which abduction is the process of generating and selecting an explanatory hypothesis that guides future inquiry (CP 5.171; 1903). Then, we introduce the contemporary concept of abduction characterized as Inference to the Best Explanation (IBE), which aim is to select a hypothesis, among a set of available hypotheses, considering their explanatory potential in terms of likelihood and loveliness (Lipton, 2004). Subsequently, we discuss IBE in relation to Bayesianism, according to which rational agents update their degrees of beliefs in a proposition based on new evidence and explanatory considerations. To illustrate our analysis, we present the software called AI-Descartes, an open-source AI system that combines logical reasoning with symbolic regression, aiming to derive scientific discovery from axiomatic knowledge and experimental data (Cornelio et al., 2023). Finally, we provide considerations about the relevance of studying abduction in the context of Artificial Intelligence.