The latest research reveals something remarkable about GPT-5: it’s beginning to act less like a sophisticated search engine and more like that brilliant colleague who helps crack tough problems. In a comprehensive study spanning mathematics, physics, biology, and computer science, researchers documented cases where the AI model didn’t just retrieve information but generated novel proofs, identified hidden connections between disparate fields, and compressed months of theoretical work into hours.
This represents a fundamental shift from AI as a tool to AI as a thinking partner, with profound implications for how scientific research will be conducted in the future.
What caught researchers’ attention was GPT-5’s ability to solve four previously unsolved mathematical problems. Not approximate solutions or suggested approaches, but actual complete solutions. One of these, Erdős Problem #848, had stumped mathematicians for decades. The AI’s contribution was a stability-style analysis that human mathematicians had overlooked, sandwiched between layers of human insight.
The real breakthrough lies in what researchers call the “compression factor.” Brian Spears from Lawrence Livermore National Laboratory used GPT-5 to model thermonuclear burn propagation in fusion experiments. Six hours of collaborative work with the AI accomplished what he estimated would have taken six person-months with a team of postdocs. This represents not just efficiency gains, but a fundamental shift in how research might be conducted.
Timothy Gowers, a Fields Medalist involved in the study, compared GPT-5’s contributions to those of a knowledgeable research supervisor: helpful, sometimes insightful, but not yet at the level where you’d list them as a co-author on most papers. This balanced perspective captures both the promise and limitations of current AI capabilities.
Perhaps the most immediately practical application emerges from GPT-5’s ability to perform “deep literature search.” This goes far beyond keyword matching to identify conceptual connections across disciplines. The model identified that a new result in density estimation was mathematically equivalent to work on “approximate Pareto sets” in multi-objective optimization, a connection the human authors had completely missed because the fields use entirely different terminology.
In another striking example, GPT-5 located solutions to 10 Erdős problems previously marked as “open,” including papers in German from decades ago. Most remarkably, the model found a solution hidden in a brief side comment between two theorems in a 1961 paper, something that had been overlooked by human reviewers for over 60 years.
This capability addresses a critical challenge in modern research: the exponential growth of scientific literature makes it increasingly difficult for researchers to stay current across relevant fields, let alone identify connections between disparate areas of study.
The research also illuminated crucial limitations that underscore the continued importance of human oversight. Derya Unutmaz’s immunology experiments showcase both the promise and the peril. GPT-5 correctly identified that 2-deoxy-D-glucose was interfering with N-linked glycosylation rather than just glycolysis in T-cells, a mechanistic insight the research team had missed despite deep expertise in the field. Yet the model also required constant human oversight to catch overconfident assertions and flawed reasoning.
Christian Coester’s work on online algorithms demonstrates another pattern: GPT-5 excels at specific, well-defined subproblems but struggles with open-ended theoretical questions. When asked to prove or disprove that a particular algorithm could achieve a certain performance bound, it produced an elegant counterexample using the Chevalley-Warning theorem. But when pushing for more general results, it often generated flawed arguments that required human correction.
A fascinating pattern emerged across disciplines: GPT-5 performs dramatically better when properly “scaffolded.” Alex Lupsasca discovered this when the model initially failed to find symmetries in black hole equations. But after working through a simpler flat-space problem first, GPT-5 successfully derived the complex curved-space symmetries, reproducing months of human work in minutes.
This scaffolding requirement reveals something fundamental about current AI capabilities. These models possess vast knowledge and computational power, but they need human expertise to direct that capability effectively. It’s like having access to a Formula 1 engine: immensely powerful, but you still need to know how to build the rest of the car and drive it.
The researchers repeatedly emphasized that using GPT-5 effectively requires deep domain expertise. You need to know when the model is hallucinating, when to push back on its assertions, and how to scaffold problems appropriately. In essence, the better you are at your field, the more value you can extract from these AI collaborators.
Not all stories in the research are triumphant. Venkatesan Guruswami and Parikshit Gopalan’s experience with “clique-avoiding codes” serves as a crucial warning. GPT-5 provided a correct proof for a problem they’d been curious about for years. Excitement turned to embarrassment when they discovered the exact same proof had been published three years earlier. The AI had essentially plagiarized without realizing it, highlighting a critical challenge for AI-assisted research: ensuring proper attribution when the model might not always identify its sources.
This incident underscores the need for robust verification processes and citation checking when working with AI systems. As these tools become more powerful and widespread, the research community will need to develop new standards and practices to maintain intellectual integrity.
The research reveals that the future of science might look less like humans versus machines and more like the best of both, working in tandem to push the boundaries of knowledge. GPT-5 isn’t just a better GPT-4; it represents a qualitative shift in capability that requires us to rethink how research is conducted.
For researchers, the message is clear: these AI systems are becoming genuine thinking partners, but they require skilled human guidance to reach their full potential. The better you are at your field, the more value you can extract from these AI collaborators. As we move forward, the question isn’t whether AI will transform scientific research, but how quickly we can adapt our workflows and practices to harness these powerful new capabilities responsibly.
The age of human-AI scientific collaboration has arrived, and those who learn to work effectively with these systems will have unprecedented advantages in advancing human knowledge.
02 December 2025
