In-Context Credit Assignment via the Core

TL;DR

This paper introduces incentive-aligned mechanisms for in-context credit assignment in AI-generated content, utilizing cooperative game theory's least core concept to fairly distribute value among creators.

Contribution

It applies the least core solution from cooperative game theory to in-context credit assignment and develops efficient algorithms for approximation.

Findings

Algorithms approximate the least core with significantly fewer LLM calls.

Approach ensures stable and fair value distribution among content creators.

Effective on web retrieval credit assignment tasks.

Abstract

We propose incentive-aligned mechanisms for in-context credit assignment: the task of assigning credit for AI-generated content (e.g. code, news articles, short-form videos) among creators whose intellectual property appears in the context window. Our approach is based on the least core solution concept from cooperative game theory, which distributes value in a way that is as stable as possible by ensuring that no subset of creators is significantly under-compensated relative to the value they could generate on their own. We develop algorithms for approximating the least core, which leverage novel routines for constraint seeding and constraint separation. On a web retrieval credit assignment task, we find that our approaches are capable of approximating the least core using orders of magnitude fewer LLM calls compared to alternative methods.