Scalable Oversight via Partitioned Human Supervision

Abstract

As artificial intelligence (AI) systems approach and surpass expert human performance across a broad range of tasks, obtaining high-quality human supervision for evaluation and training becomes increasingly challenging. Our focus is on tasks that require deep knowledge and skills of multiple domains. Unfortunately, even the best human experts are knowledgeable only in a single narrow area, and will not be able to evaluate the correctness of advanced AI systems on such superhuman tasks. However, based on their narrow expertise, humans may provide a weak signal, i.e., a complementary label indicating an option that is incorrect. For example, a cardiologist could state that "this is not related to cardiology,'' even if they cannot identify the true disease. Based on this weak signal, we propose a scalable oversight framework that enables us to evaluate frontier AI systems without the need to prepare the ground truth. We derive an unbiased estimator of top-1 accuracy from complementary labels and quantify how many complementary labels are needed to match the variance of ordinary labels. We further introduce two estimators to combine scarce ordinary labels with abundant complementary labels. We provide finite-sample deviation guarantees for both complementary-only and the mixed estimators. Empirically, we show that we can evaluate the output of large language models without the ground truth, if we have complementary labels. We further show that we can train an AI system with such weak signals: we show how we can design an agentic AI system automatically that can perform better with this partitioned human supervision. Our code is available at https://github.com/R-Yin-217/Scalable-Oversight-via-Human-Partitioned-Supervision.

Summary

This paper proposes using 'complementary labels'—where domain experts indicate incorrect rather than correct options—as a scalable oversight mechanism for evaluating and training AI systems that exceed expert human performance. The authors derive unbiased estimators of top-1 accuracy from these weak signals and provide finite-sample guarantees, demonstrating that frontier LLMs can be evaluated and fine-tuned without ground truth labels. They also show that automatically designed agentic systems can leverage this partitioned supervision to improve performance.

Key Points

• •Complementary labels (indicating wrong answers) can be provided by domain experts outside their narrow specialty, enabling oversight of superhuman AI systems.
• •Authors derive an unbiased estimator of top-1 accuracy from complementary labels and quantify sample requirements compared to ordinary labels.
• •Two mixed estimators combine scarce ordinary labels with abundant complementary labels, with finite-sample deviation guarantees provided for both.
• •Empirically validated on frontier LLMs: models can be evaluated without ground truth and trained/fine-tuned using only weak complementary supervision.
• •Agentic AI systems can be automatically designed to leverage partitioned human supervision, improving performance in superhuman task regimes.