Validity Threats for Foundation Model Research

2026-06-03

AI Summary

This study addresses the widespread reliance on low-cost proxies—such as surrogate experiments, observational studies, or single training runs—in foundational model research due to prohibitive computational costs. While pragmatic, these approaches entail implicit threats to validity that can undermine the reliability of empirical conclusions. For the first time, this work systematically adapts validity theory from the social sciences to this domain, employing a causal inference framework to dissect inherent limitations across four dimensions: statistical, internal, external, and construct validity. The analysis uncovers long-overlooked validity risks in the literature, clarifies trade-offs among alternative methodological strategies, and offers a practical toolkit for assessing validity. This contribution provides actionable guidance for designing more rigorous and trustworthy studies in foundational model research.

Abstract

Controlled experiments are the backbone of machine learning research, but at the scale of modern foundation models, they have become prohibitively expensive. Instead, the community increasingly relies on research strategies that approximate the ideal experiment at a fraction of the cost: proxy experiments and scaling laws, observational studies with publicly available models, and single-run designs that leverage variation within individual training runs. In this work, we argue that there is no free lunch when approximating large-scale experiments on a compute budget. Specifically, savings in compute come at the cost of validity threats -- hidden and sometimes untestable assumptions that, when violated, can invalidate research claims. To help navigate such threats, we propose an evaluation framework that casts foundation model research as a causal inference problem. Within this framework, we evaluate different research strategies through four types of validity adapted from the empirical social sciences -- statistical, internal, external, and construct validity. We find that each strategy comes with a characteristic validity profile: proxy experiments trade external and construct validity for statistical and internal validity; observational studies face confounding and effect heterogeneity; and single-run designs are strained by interference between treated units. This analysis reveals several validity threats that have received insufficient attention in the literature. Overall, our evaluation framework provides researchers with a practical toolkit for scrutinizing validity threats in foundation model research~designs.

Problem

• validity threats
• foundation models
• causal inference
• proxy experiments
• observational studies

Innovation

• validity threats
• causal inference
• foundation models
• research design
• scaling laws