BROWSE
Detail
Journal Article
AI 파운데이션 모델 중심 인공지능 연구 동향의 변화
- Issue Date
- 2025-12
- Citation
- 전자통신동향분석, v.40, no.6, pp.1-14
- ISSN
- 1225-6455
- Publisher
- 한국전자통신연구원
- Language
- Korean
- Type
- Journal Article
- Abstract
- Recent advances in artificial intelligence (AI) have revealed a shift in research practices. Researchers are now constructing shared foundation models with strong generalization and emergent capabilities instead of independently developing domain-specific models. This study surveys the transformations driven by this shift, outlines their conceptual backgrounds, introduces two main approaches to leverage them, states the key structural limitations and proposed remedies, and presents the resulting changes in training, data, and evaluation paradigms. Rather than listing the advances, we emphasize the motivations and implications behind them, offering perspectives on how foundation models reshape the research ecosystem and inform future directions in AI.
- Keyword
- Agentic AI, Benchmark, CoT, Datacentric AI, DPO, Foundation Model, ICL, Instruction Tuning, LLM, LoRA, Multimodal LLM, RAG, RLHF, RLAIF, VLA, World Model
- KSP Keywords
- Domain-specific, World model, artificial intelligence, future directions
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)
