The role of LLMs in theory building

2025/01/01 by Aníbal Monasterio Astobiza · 1 voice
Social Sciences · Computer Science · Biochemistry, Genetics and Molecular Biology · #Artificial Intelligence in Law #Natural Language Processing Techniques #Biomedical Text Mining and Ontologies

paper · doi:10.1016/j.ssaho.2025.101617

openalex publication_date 2025/01/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/23

Abstract

Large linguistic models, such as GPT-3.5 and subsequent versions (e.g. GPT-4 or GPT - 4o ), have shown impressive abilities in generating human-like text and performing a variety of natural language processing tasks. However, a fundamental question in the field of artificial intelligence is whether these models can truly represent meaning and assist scientists in building scientific theories. This paper aims to address this question by conducting a thorough conceptual analysis of existing large linguistic models and their capabilities in representing and reasoning about meaning for the purpose of theory building. My conclusions suggest that while these models have made significant progress in representing and manipulating language, they still face limitations in their ability to represent abstract and complex concepts, and the application of these models in building scientific theories should be guided by specific research questions and informed hypotheses that can be tested and developed into robust theories.

Citations

• Efficient Estimation of Word Representations in Vector Space
• Estimating the reproducibility of psychological science
• Artificial intelligence: a modern approach
• Cognitive Load During Problem Solving: Effects on Learning
• Human-level concept learning through probabilistic program induction
• Distributional Structure
• The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence
• Algorithmic injustice: a relational ethics approach
• Training language models to follow instructions with human feedback
• Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity
• Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
• Aion Framework: Dimensional Emergence of AI Consciousness, Observer-Induced Collapse, and Cosmological Portal Dynamics
• Scaling Laws for Reward Model Overoptimization
• The debate over understanding in AI’s large language models
• Scientific discovery in the age of artificial intelligence

Discussions

• Read the full article here: doi.org/10.1016/j.ss... [bsky, 0 points, 0 comments]

Related

• The Quo Vadis of the Relationship between Language and Large Language Models
• Large Language Models as Proxies for Theories of Human Linguistic Cognition
• A Perspective on Large Language Models, Intelligent Machines, and Knowledge Acquisition
• Talking About Large Language Models
• "Understanding AI": Semantic Grounding in Large Language Models