Gov't research institutes launch 'science AI' push, from AI agents to autonomous labs

A project to boost the speed and productivity of scientific research by combining AI with decades of research data and expertise accumulated by government-funded institutes is now getting underway.

The National AI for Science Center, or NAIS, under the National Research Council of Science and Technology (NST) announced Tuesday it is accepting applications for new seed-stage projects under its 2026 NAIS AI Convergence Research Program, with a deadline of Oct. 20.

The program pairs research data, infrastructure and domain expertise held by government-funded institutes with AI capabilities from companies, universities and research institutions. The goal is to secure "AI-ready research assets" that can be used repeatedly across scientific and technological research settings.

Such assets include domain-specific AI agents, datasets refined, structured and standardized for AI training, and operational protocols for physical AI and autonomous laboratory systems.

The program follows a two-stage structure — seed stage leading to a demonstration stage. In the seed stage, teams carry out proof-of-concept work on AI research assets. About 15 projects will be selected, each receiving up to 200 million won ($148,000) over one year.

Projects that show promise will advance to the follow-on demonstration stage. Up to 30 percent of seed-stage selections will be eligible for 500 million to 1 billion won per project over one year, with the aim of developing and validating technologies to a level ready for use in actual research settings. The availability and scale of demonstration-stage funding may be adjusted depending on the following year's budget.

Lead institutions must be NST-affiliated research institutes or their subsidiary bodies. At least one industry, academic or research partner must participate as a joint research institution.

Selected research teams will receive mentoring from AI specialists as well as computing resources including GPUs. A research environment with access to shared AI models, large language models and APIs will also be provided.

AI research assets produced during the projects will be registered on AI-OS, an integrated science AI platform being built by NAIS. Findings that can be made public are to be shared across the broader scientific and technological research community.

Yoo Yong-gyun, director of NAIS, said the center would "create a virtuous cycle in which outstanding seed-stage projects with proven potential are linked to the demonstration stage, and AI-ready research assets accumulate and are put to use through AI-OS."

Kim Young-sik, chairman of NST, said that "as AI transforms the speed and methods of scientific research, it is important to convert the domain expertise accumulated by government-funded institutes into reusable AI research assets," adding that the council would "bring together the capabilities of research institutes and industry-academic-research partners to help outstanding results spread across the broader research community."