BROWSE
Detail
Journal Article
AI Scientist 기술 개발 동향
- Issue Date
- 2026-10
- Citation
- 전자통신동향분석, v.41, no.5, pp.72-84
- ISSN
- 1225-6455
- Publisher
- 한국전자통신연구원
- Language
- Korean
- Type
- Journal Article
- Abstract
- Advances in foundation models, scientific agents, multimodal reasoning, automated experimentation, and laboratory robotics enable artificial intelligence (AI) systems to participate in the scientific discovery cycle. However, capabilities remain fragmented, and no common framework unifies reasoning, experimentation, validation, and collaboration. This article defines an AI Scientist as an agentic system that acquires knowledge, prioritizes hypotheses, conducts digital or physical experiments, analyzes results, and refines research plans. It proposes capability levels spanning AI Co-Scientists, autonomous AI Scientists, and AI research organizations; reviews international and Korean systems across scientific domains; and presents a human–AI collaborative reference architecture. The analysis shows strong performance in literature synthesis, hypothesis generation, code-based experimentation, and closed-loop laboratory automation, but limits in long-horizon coherence, reproducibility, and physical-world reliability. These findings support priorities in human–AI collaboration, evidence-centered evaluation, standardized tool interfaces, and staged domain validation.
- Keyword
- AI Co-Scientist, AI for Science, AI Scientist, Autonomous Scientific Discovery, Self-Driving Laboratory
- KSP Keywords
- Closed-loop, Collaborative Reference Architecture, Laboratory automation, Literature synthesis, Multimodal reasoning, Reference Architecture(RA), Scientific discovery, Self-Driving, artificial intelligence, hypothesis generation, physical experiment
This work is distributed under the term of Korea Open Government License (KOGL)
(Type 4: : Type 1 + Commercial Use Prohibition+Change Prohibition)
(Type 4: : Type 1 + Commercial Use Prohibition+Change Prohibition)
