How Open Must Language Models be to Enable Reliable Scientific Inference?

TL;DR

This paper examines how the openness of language models affects the reliability of scientific inference, highlighting issues with closed models and proposing guidelines for responsible model use in research.

Contribution

It analyzes the impact of model openness on scientific inference and offers recommendations for mitigating related risks in research practices.

Findings

Closed models pose threats to reliable inference

Open models can improve scientific research validity

Guidelines for responsible model use are proposed

Abstract

How does the extent to which a model is open or closed impact the scientific inferences that can be drawn from research that involves it? In this paper, we analyze how restrictions on information about model construction and deployment threaten reliable inference. We argue that current closed models are generally ill-suited for scientific purposes, with some notable exceptions, and discuss ways in which the issues they present to reliable inference can be resolved or mitigated. We recommend that when models are used in research, potential threats to inference should be systematically identified along with the steps taken to mitigate them, and that specific justifications for model selection should be provided.