OpenAI: AI-агенты для кода ускоряют научные вычисления и открытия в геномике

OpenAI published a field report on July 30, 2026, about how scientists are using AI coding agents to modernize scientific computing and accelerate software development and discoveries in genomics and related fields.

What OpenAI's report describes

OpenAI's report documents real-world practice: researchers are increasingly applying agentic agents — AI that independently reads, writes, runs, and fixes code — to update scientific software. The field report format is a look at how the technology works "in the field," not an announcement of a new model: OpenAI gathers the experience of scientific teams and shows where AI agents are already delivering results.

This isn't about one-off suggestions in an editor, but about the agent's full participation in the development cycle — from parsing unfamiliar code to writing and testing new modules.

Key facts from the report:

• Publisher — OpenAI, format — field report, published July 30, 2026
• Topic — the use of AI coding agents in scientific computing
• Example field — genomics, as well as related scientific disciplines
• Effect — accelerated software development and scientific discovery
• Audience — researchers and scientific developers

How agents are changing scientists' work

AI agents are taking over engineering routine work that previously took time away from research. According to OpenAI's observations, coding agents help parse unfamiliar and outdated code, port it to modern tools, write tests, and automate data processing.

This matters because scientific software is among the hardest to maintain. A significant share of lab code was written over years by domain specialists rather than professional engineers, accumulating technical debt. An agent that understands such code and can safely modify it removes a barrier that has held back tool upgrades for years.

Why does this matter for science?

Genomics is a telling example in OpenAI's report: data volumes and the complexity of computational pipelines are especially high here, so faster code development translates directly into a faster pace of scientific discovery. The quicker a team updates and runs its software, the quicker it can test hypotheses.

OpenAI emphasizes that the effect is not limited to one discipline — the approach applies to scientific computing broadly, "in genomics and beyond."

"Scientists are using AI coding agents to modernize scientific computing and accelerate discovery in genomics and beyond,"

OpenAI's field report states.

What this means

Agentic AI is moving beyond chat assistants and product development into fundamental science: coding agents are becoming a working tool for researchers, shortening the path from idea to result. For fields like genomics, where the bottleneck isn't hypotheses but data and code engineering, this could speed up the very pace of discovery. If OpenAI's report reflects a durable trend, in the coming years scientific teams will increasingly assemble and rewrite their software together with AI agents rather than by hand.

Frequently asked questions

What are AI coding agents?

These are AI tools that don't just suggest lines of code but independently read, write, run, and fix programs within a given task. In OpenAI's report, such agents are applied to scientific software.

Which scientific fields does the report cover?

OpenAI names genomics as a key example and notes that the approach extends to other scientific computing as well — in the report's words, "in genomics and beyond."