OpenAI's New Push: Transforming Science with AI

Revolutionizing Scientific Discovery

In the last few years, OpenAI's large language models (LLMs) have been making waves in various sectors, and now, the company is setting its sights on the scientific community. OpenAI for Science, a new team dedicated to leveraging LLMs for scientific research, was launched in October 2025.

Scientists from diverse fields, such as mathematics, physics, and biology, have shared experiences of how LLMs, particularly GPT-5, have aided their research, leading to discoveries or nudging them towards solutions they might have missed.

A Late Entrant, But with Ambitions

While OpenAI is entering the AI-for-science arena relatively late, with Google DeepMind having had an AI team for several years, the company aims to make up for lost time. OpenAI's Vice President, Kevin Weil, leading the OpenAI for Science team, emphasizes the potential impact of future versions of this technology on science, promising new medicines, materials, and devices.

Accelerating Science: The North East and Indian Perspective

The potential implications of this development extend beyond the United States and could significantly impact the scientific community in North East India and the broader Indian context. Collaborative research projects between Indian institutions and OpenAI could lead to groundbreaking discoveries and technological advancements, further enhancing India's global standing in scientific research.

The Journey So Far: From SAT Scores to Unsolved Problems

OpenAI has been pushing the boundaries of what LLMs can do since the announcement of its first reasoning model, o1, in December 2024. Initially, LLMs were impressive for their ability to excel in tasks such as acing math competitions and solving graduate-level physics problems. However, with the introduction of GPT-5, the models have become far better at solving complicated problems, thanks to the inclusion of a reasoning model.

The Hype and the Reality

While the excitement surrounding GPT-5's capabilities is evident, there are also concerns about overhyping the technology's potential. In October 2025, senior figures at OpenAI claimed that GPT-5 had solved several unsolved math problems. However, mathematicians quickly pointed out that the model had merely dug up existing solutions in old research papers.

Kevin Weil acknowledges that finding solutions that already exist but have been forgotten can still be a significant acceleration. He downplays the idea that LLMs are on the verge of making game-changing new discoveries, stating, "I don't think models are there yet. Maybe they'll get there. I'm optimistic that they will. But our mission is to accelerate science."

Real-world Impact: Case Studies from Scientists

In November 2025, OpenAI published a series of anecdotal case studies contributed by scientists, both within and outside the company, illustrating how GPT-5 has been used in their research. Most of the cases involve scientists using GPT-5 to find references and connections to existing work, sketch mathematical proofs, and suggest ways to test hypotheses in the lab.

The Future of Science with AI

As LLMs continue to improve, scientists are increasingly recognizing their potential as valuable tools for scientific research. Many believe that all scientists will be using LLMs in the near future. However, there are also concerns about the technology's limitations, such as the potential for errors and the need for caution when interpreting its output.

Reflections and Looking Forward

OpenAI's push into the AI-for-science sector represents an exciting new frontier in the ongoing development of AI. While there are challenges to be addressed, the potential for LLMs to accelerate scientific research and discovery is undeniable. As the technology continues to evolve, it will be fascinating to see how it shapes the future of science, both in North East India and globally.