Abstract: Recent advances in large language models (LLMs) are remarkable. LLMs can answer most questions with humanlike fluency and generate useful sentences and paragraphs in just a few seconds. At the same time, NLP researchers have raised serious concerns about LLMs, including the frequent occurrences of hallucination and toxic responses. I raise another serious concern, and that is the lack of cultural intelligence. These LLMs produce responses that represent the most prevalent culture of their training datasets, unable to produce culturally sensitive responses. In my talk, I will present some research toward achieving cultural intelligence of LLMs. This research involves first analyzing and quantifying the cultural intelligence of current LLMs, collecting and annotating datasets representing different cultures, and fine tuning models with those datasets so that they acquire some cultural intelligence.
Bio: Alice Oh is a Professor in the School of Computing at KAIST. She received her MS in 2000 from Carnegie Mellon University and PhD in 2008 from MIT. Her major research area is at the intersection of natural language processing (NLP) and computational social science. She collaborates with social scientists to study topics such as political science, education, and history, developing NLP models for various textual data including legislative bills, historical documents, news articles, social media posts, and personal conversations. She has served as a Tutorial Chair for NeurIPS 2019, Diversity & Inclusion Chair for ICLR 2019, Program Chair for ICLR 2021, Senior Program Chair for NeurIPS 2022, and General Chair for ACM FAccT 2022. She is serving as a General Chair for NeurIPS 2023.
Abstract: A truly ubiquitous environment is where human-machine interactions are intuitive, reliable, and compatible. This requires an intelligent platform that is versatile and adaptable to evolving tasks and dynamic environments. While there are extensive efforts in addressing this challenge through massive data and learning algorithms, there is yet to be a cohesive solution to improve the actual physical interactions. Recent developments in soft robots with their unconventional material-based solutions and modular robots with a multitude of configurations propose possible avenues to extend the capacities of robotics. This talk will highlight the recent progress in soft-material robots and reconfigurable origami robots that aim at achieving comprehensive solutions toward diverse “softer” human-robot applications.
Bio: Prof. Jamie Paik is director and founder of Reconfigurable Robotics Lab (RRL) of Swiss Federal Institute of Technology (EPFL) and a core member of Swiss National Centers of Competence in Research (NCCR) Robotics consortium. RRL’s research leverages expertise in design and advanced manufacturing toward reconfigurable robotic platforms that push the physical limits of material and mechanisms. Her latest research effort is in soft robotics and self-morphing Robogami (robotic origami). Robogami transforms autonomously its planar shape to 2D or 3D by folding like the paper art, origami. Soft material robots and robogamis are designed to be interactive with the users and their environments through both innate and active reconfigurations. Such characteristics of the RRL’s robots have direct applications in medical, automobile, space, communication, and wearable robots. While this novel technology has been published in multiple academic journals such as in Soft Robotics Journal, IEEE Transactions in Robotics, Nature, and Science, RRL’s spin-offs, MIROS and Foldaway-Haptics, have pushed the boundaries of the industrial applications of these robots as seen in TED conferences 2019 and 2023. One of the robogamis was a part of Mercedez’s 2020 concept car, AVTR, presented during CES 2020, and MIROS in CES 2023.