Training Data Attribution (TDA): Examining Its Adoption & Use Cases

TL;DR

This paper explores the development, adoption, and societal implications of Training Data Attribution (TDA) tools for AI, emphasizing their potential to reduce risks and the challenges related to data disclosure.

Contribution

It analyzes the feasibility of creating effective TDA tools for large-scale LLMs and discusses policy and societal impacts of TDA adoption.

Findings

TDA tools could significantly reduce AI risks if developed effectively.

Willingness of AI labs to disclose training data is a key bottleneck.

TDA has potential to enable new policies and safety measures.

Abstract

This report investigates Training Data Attribution (TDA) and its potential importance to and tractability for reducing extreme risks from AI. First, we discuss the plausibility and amount of effort it would take to bring existing TDA research efforts from their current state, to an efficient and accurate tool for TDA inference that can be run on frontier-scale LLMs. Next, we discuss the numerous research benefits AI labs will expect to see from using such TDA tooling. Then, we discuss a key outstanding bottleneck that would limit such TDA tooling from being accessible publicly: AI labs' willingness to disclose their training data. We suggest ways AI labs may work around these limitations, and discuss the willingness of governments to mandate such access. Assuming that AI labs willingly provide access to TDA inference, we then discuss what high-level societal benefits you might see. We list and discuss a series of policies and systems that may be enabled by TDA. Finally, we present an evaluation of TDA's potential impact on mitigating large-scale risks from AI systems.