Test-time Computing: from System-1 Thinking to System-2 Thinking

2025-01-05

AI Summary

This paper addresses the tension between computational efficiency and cognitive capability enhancement during large language model (LLM) inference. We propose the first unified test-time computation framework that explicitly models the System-1 (intuitive) to System-2 (deliberative) transition. Methodologically, we systematically integrate seven complementary techniques—including parameter updating, latent-space editing, self-correction, and tree search—characterizing their hierarchical roles and synergistic mechanisms in cognitive upgrading. Our contributions are threefold: (1) we formally define and empirically validate a cognitive evolution pathway for test-time computation, demonstrating consistent gains in out-of-distribution robustness and complex problem-solving; (2) we identify critical bottlenecks hindering the shift from weak to strong System-2 behavior—particularly dynamic computational allocation and interpretable calibration; and (3) we outline concrete research directions toward scalable, theoretically grounded “reasoning-as-refinement” architectures.

Technology Category

• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

The remarkable performance of the o1 model in complex reasoning demonstrates that test-time computing scaling can further unlock the model's potential, enabling powerful System-2 thinking. However, there is still a lack of comprehensive surveys for test-time computing scaling. We trace the concept of test-time computing back to System-1 models. In System-1 models, test-time computing addresses distribution shifts and improves robustness and generalization through parameter updating, input modification, representation editing, and output calibration. In System-2 models, it enhances the model's reasoning ability to solve complex problems through repeated sampling, self-correction, and tree search. We organize this survey according to the trend of System-1 to System-2 thinking, highlighting the key role of test-time computing in the transition from System-1 models to weak System-2 models, and then to strong System-2 models. We also point out a few possible future directions.

Problem

• Computational Efficiency
• Model Robustness
• Complex Problem Solving

Innovation

• Computational Increase
• Model Performance Enhancement
• Complexity-based Strategy