Uploaded May 2024 | Updated September 2026, 1 day ago
Abstract: The advancement of science, engineering, and design depends on scientists' cognitive abilities to innovate beyond existing ideas. While human cognition excels at detecting patterns and forming original ideas, it is also hampered by cognitive biases and limitations, such as working memory and processing bandwidths. These limitations can lead to a detrimental fixation on suboptimal ideas and paths. With the exponential increase in new information, navigating and sifting through this deluge to derive useful insights becomes even more daunting.
Recent advances in large language models (LLMs) present a significant opportunity for addressing these challenges by augmenting scientists across several stages of their science-ideation workflows. These stages range from discovering relevant literature and reading important papers within it, to synthesizing insights and ultimately generating novel ideas. In this talk, I will present my prior work on building computational systems to augment scientists and their discovery-synthesis-insight stages. Furthermore, I will introduce how future work may more tightly integrate LLMs and system design to enhance scientists' cognitive creativity and efficiency.
Bio: Hyeonsu Kang is a Computer Science Ph.D. candidate at the Human-Computer Interaction Institute at Carnegie Mellon University, advised by Niki Kittur. His work blends AI with interaction design to innovate the ways in which we engage with AI, aiming to boost creative thinking and cognitive productivity. He has developed systems that help users think outside the box, find and synthesize existing knowledge, and benefit from the exchange of feedback among peers, crowds, and experts. His approach is grounded in cognitive science, examining how we apply concepts across domains to make analogies conducive to new inspirations. Kang was a research intern for two summers at Semantic Scholar. His work has been applied in real-world settings, such as allocating two million dollars in prize money for innovation projects at Conservation X Labs. He obtained his BS from Seoul National University.
Abstract: The advancement of science, engineering, and design depends on scientists' cognitive abilities to innovate beyond existing ideas. While human cognition excels at detecting patterns and forming original ideas, it is also hampered by cognitive biases and limitations, such as working memory and processing bandwidths. These limitations can lead to a detrimental fixation on suboptimal ideas and paths. With the exponential increase in new information, navigating and sifting through this deluge to derive useful insights becomes even more daunting.
Recent advances in large language models (LLMs) present a significant opportunity for addressing these challenges by augmenting scientists across several stages of their science-ideation workflows. These stages range from discovering relevant literature and reading important papers within it, to synthesizing insights and ultimately generating novel ideas. In this talk, I will present my prior work on building computational systems to augment scientists and their discovery-synthesis-insight stages. Furthermore, I will introduce how future work may more tightly integrate LLMs and system design to enhance scientists' cognitive creativity and efficiency.
Bio: Hyeonsu Kang is a Computer Science Ph.D. candidate at the Human-Computer Interaction Institute at Carnegie Mellon University, advised by Niki Kittur. His work blends AI with interaction design to innovate the ways in which we engage with AI, aiming to boost creative thinking and cognitive productivity. He has developed systems that help users think outside the box, find and synthesize existing knowledge, and benefit from the exchange of feedback among peers, crowds, and experts. His approach is grounded in cognitive science, examining how we apply concepts across domains to make analogies conducive to new inspirations. Kang was a research intern for two summers at Semantic Scholar. His work has been applied in real-world settings, such as allocating two million dollars in prize money for innovation projects at Conservation X Labs. He obtained his BS from Seoul National University.
