Realist and Pluralist Conceptions of Intelligence and Their Implications on AI Research

TL;DR

The paper identifies a fundamental split in AI research between Intelligence Realism (a single universal cognitive core) and Intelligence Pluralism (multiple context-dependent intelligences). It argues that these hidden commitments mold methodological choices, interpretation of results (including scaling laws and emergent capabilities), and AI risk governance. By making these assumptions explicit, the authors offer a vocabulary to distinguish genuinely empirical disagreements from philosophical ones and to navigate the spectrum of intermediate positions. The work emphasizes a continuum rather than a binary view and argues for context-sensitive safety and governance informed by the underlying conception of intelligence.

Abstract

In this paper, we argue that current AI research operates on a spectrum between two different underlying conceptions of intelligence: Intelligence Realism, which holds that intelligence represents a single, universal capacity measurable across all systems, and Intelligence Pluralism, which views intelligence as diverse, context-dependent capacities that cannot be reduced to a single universal measure. Through an analysis of current debates in AI research, we demonstrate how the conceptions remain largely implicit yet fundamentally shape how empirical evidence gets interpreted across a wide range of areas. These underlying views generate fundamentally different research approaches across three areas. Methodologically, they produce different approaches to model selection, benchmark design, and experimental validation. Interpretively, they lead to contradictory readings of the same empirical phenomena, from capability emergence to system limitations. Regarding AI risk, they generate categorically different assessments: realists view superintelligence as the primary risk and search for unified alignment solutions, while pluralists see diverse threats across different domains requiring context-specific solutions. We argue that making explicit these underlying assumptions can contribute to a clearer understanding of disagreements in AI research.