Center for Social Information Sciences (CSIS) Seminar

Abstract: While the success of large language models (LLMs) increases demand for machine-generated text, current pay-per-token pricing schemes create a misalignment of incentives known as moral hazard: Providers of text-generation services have an incentive to cut costs by preferring a cheaper model over the cutting-edge one, and this can be done "behind the scenes" since the provider performs inference internally. In this work, we approach this issue from an economic perspective by proposing a pay-for-performance, contract-based framework for incentivizing quality. We study a principal-agent game where the agent generates text using costly inference, and a contract determines the principal's payment for the text according to an automated quality evaluation. Since standard contract theory is inapplicable when internal inference costs are uncertain, we introduce cost-robust contracts. As our main theoretical contribution, we characterize optimal cost-robust contracts through a direct correspondence to optimal composite hypothesis tests from statistics. We evaluate our framework empirically by deriving contracts for a range of objectives and LLM evaluation benchmarks, and find that cost-robust contracts sacrifice only a marginal amount of objective value compared to their cost-aware counterparts.

Joint work with Inbal Talgam-Cohen and Ohad Einav.