Early science acceleration experiments with GPT-5

2025-11-20

AI Summary

The scientific community lacks a systematic understanding of frontier AI models—such as GPT-5—in augmenting interdisciplinary research. Method: We propose a human–AI collaborative research paradigm, leveraging GPT-5 as an interactive reasoning engine that tightly integrates domain-expert judgment with AI-driven generative inference across six disciplines: mathematics, physics, astronomy, computer science, biology, and materials science. Contribution/Results: First, we empirically demonstrate GPT-5’s capability to assist in resolving four long-standing open problems in mathematics. Second, we introduce the reproducible AI-Augmented Research Protocol (AARP), which establishes a closed-loop workflow spanning problem modeling, hypothesis generation, formal verification, and result interpretation. Experiments show substantial gains in exploratory efficiency and cross-domain transferability, while reaffirming the irreplaceable roles of human experts in problem formulation, logical validation, and scientific interpretation.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Knowledge Representation and Reasoning: Automated Reasoning and Theorem Proving
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

AI models like GPT-5 are an increasingly valuable tool for scientists, but many remain unaware of the capabilities of frontier AI. We present a collection of short case studies in which GPT-5 produced new, concrete steps in ongoing research across mathematics, physics, astronomy, computer science, biology, and materials science. In these examples, the authors highlight how AI accelerated their work, and where it fell short; where expert time was saved, and where human input was still key. We document the interactions of the human authors with GPT-5, as guiding examples of fruitful collaboration with AI. Of note, this paper includes four new results in mathematics (carefully verified by the human authors), underscoring how GPT-5 can help human mathematicians settle previously unsolved problems. These contributions are modest in scope but profound in implication, given the rate at which frontier AI is progressing.

Problem

• Demonstrating GPT-5's capabilities for scientific research acceleration across multiple disciplines
• Identifying where AI accelerates work versus where human input remains essential
• Providing verified examples of GPT-5 helping solve previously unsolved mathematical problems

Innovation

• GPT-5 accelerates research across multiple scientific fields
• Human-AI collaboration produces verified mathematical results
• Case studies demonstrate frontier AI capabilities and limitations