Morals and Reasoning: Formalizing Moral Influence on Reasoning and AI Systems Alignment

Published: 24 Jul 2025, Last Modified: 01 Aug 2025Social Sim'25EveryoneRevisionsBibTeXCC BY 4.0
Keywords: Moral Values, Ethical Reasoning, Social Dynamics, AI Alignment, Institutional Sustainability, Values Diffusion Effect, Generative Agent-Based Models, Concordia Framework, Psychoanalytical Formalization, Social Network Analysis
TL;DR: This paper formalizes how moral values enhance reasoning, social dynamics, and how it can be adapted for AI alignment, introducing the Values Diffusion Effect.
Abstract: Reasoning frameworks are built upon ethical principles or the lack thereof. This paper provides a psychoanalytical perspective on reasoning patterns and dynamics, focusing on the impact of moral values on decision-making and social outcomes. We formalize how moral values shape perception and enhance institutional performance, introducing the Values DIffusion Effect to describe their positive, spreading influence. We provide a systematic formalization for implementing ethical evaluation, offering a standard and replicable process for incorporating ethically guided decision-making. We examine how institutions can sustainably evolve by effectively considering dynamic external conditions through efficient decision-making systems and social interactions. For example, in economic institutions, systems and processes can evolve through ethical consideration of external factors to achieve a sustainable state, addressing issues like unbounded economic objectives, which should be tied to market satisfaction and sustainability. In artificial intelligence (AI), we propose technical considerations for AI alignment that build on findings from the sociological domain regarding ethically grounded reasoning. Simulation strategies are also explored, leveraging the ability of autoregressive models to accurately estimate social contexts within generative agent-based models (GABMs) using frameworks such as Concordia to measure efficiency gains from ethically oriented reasoning frameworks. Applications from these findings include sustainable economic institutions and ethically aligned AI systems, addressing challenges like resource allocation and AI decision-making.
Submission Number: 26
Loading