In a student-centered teaching project, undergraduates analyzed botanical elements in Taylor Swift music videos to activate prior knowledge and reinforce complex plant science concepts. The experience reports improved comprehension, high student satisfaction, and potential to counter “plant blindness” through popular culture.

A research team from Merck (China) has proposed Jigsaw-LightRAG, a novel method that improves how artificial intelligence systems manage and update knowledge graphs.  It decomposes a global knowledge graph (KG) into per-document subgraphs and leverages document lifecycle states (New, Modified, Persistent, Deleted) so that LLM-based extraction is invoked only for changed documents. The updated global KG is then reconstructed via code-level aggregation and strict string-matching deduplication without additional LLM token consumption. Experiments across public benchmarks (LongBench, PubMedQA, and QASPER) demonstrate that in localized corpus changing scenarios, LLM token consumption drops by orders of magnitude (with DELETE operations requiring 0 tokens), while KG structural integrity remains stable and question-answering performance stays on par with full-rebuild baselines.

A research team led by Professor Taesung Kim from the School of Mechanical Engineering has developed a technology that precisely controls the internal structure of semiconductors using heat, much like stamping out “bungeoppang” (fish-shaped pastry) in a mold. The team reported that this approach improves the performance of next-generation artificial intelligence (AI) hardware. With this technology, complex AI computations can be processed more quickly using significantly less electricity than before.

Korean researchers have achieved a breakthrough in the safety of generative AI. They developed a vision language model optimized for safety and released it for the first time. In this model, AI can preemptively analyze both images and text, even detecting risks. The research team is paving the way for a safe AI era. ETRI announced that it has unveiled a new type of vision language model called “Safe LLaVA,” which structurally enhances safety in generative AI models.

Natural language processing (NLP), a subfield of artificial intelligence that uses computational techniques to interpret, analyze, and generate human language, encompasses a range of tasks and techniques. These include the large language models (LLMs) that power chatbots and other types of systems, as well as specific approaches (some employed by LLMs), including information extraction and text mining.