Symposium on Ethics & Epistemology in Artificial Intelligence

Symposium on Ethics & Epistemology in Artificial Intelligence

Speakers Steven Gouveia (University of Porto) y Mariana Olezza (UBA/CIECE)

Abstract:

Medical AI and Abduction & Effective Altruism framework: In this talk, we argue that addressing the “Trust Gap” in AI requires a focus on explanations from an Explainable AI (xAI) perspective. Current xAI frameworks fall short, so we propose a pragmatist turn that draws on how explanations function among human agents. Human explanations are typically (i) social and (ii) abductive, providing understanding by answering contrastive why-questions: “Why did X happen instead of Y?” (Miller, 2019). Testing this notion in AI, we examine the capacity of different systems for social and abductive reasoning. In particular, Large Language Models (LLMs) and transformer architectures show notable potential to engage in abductive reasoning. Leveraging these capabilities could improve the interpretability of AI outputs and enhance trust between AI systems and human users, helping bridge the Trust Gap across a variety of domains.

Mariana Olezza (UBA/CIECE)

Abstract:

Continuity and Rupture: The (Non)Convergence of Scientific Paradigms in Computing and Economics: This presentation reflects on Peirce’s fallibilism (Hetherington, 2025) and its implications for understanding scientific convergence. While Peirce’s synechism conceives the universe as a continuum, fallibilism (Burch, 2021) reminds us that any metaphysical or epistemological stance remains provisional. Within this framework, two sciences illustrate divergent tendencies: computing and economics. Early computing was grounded in determinism—logical machines and deductive systems (Peirce, 1887). Yet, with the emergence of Artificial Neural Networks (ANNs), Big Data, Generative Adversarial Networks (GANs) (Goodfellow et al., 2014), Large Language Models (LLMs) (Vaswani et al., 2017), questions arise about whether deterministic machines can replicate human cognition. Some researchers uphold determinism; others see cognition as an indeterminate, self-organizing phenomenon. Similarly, in economics, classical and neoclassical models presuppose determinism—perfect rationality and convergence to equilibrium—whereas Keynesian and post-Keynesian traditions embrace indeterminism, emphasizing uncertainty, history, and discretion.

Moderador Javier Legris (UBA/IIEP-CONICET)

Lugar | FCE UBA

Organiza CIECE – Research Center in Epistemology of Economic Sciences

Language: English