UROP Project
Can AI Interpret Nuanced Expression?
Artificial Intelligence, Large Language Models, Natural Language Processing
Research Mentor: Dr. Rashad Aziz, He/Him
Department, College, Affiliation: Office of the Provost, N/A
Contact Email: raziz@admin.fsu.edu
Research Assistant Supervisor (if different from mentor):
Research Assistant Supervisor Email:
Faculty Collaborators: Dr. Solveig Brown She/Her
Faculty Collaborators Email: sbrown7@fsu.edu
Department, College, Affiliation: Office of the Provost, N/A
Contact Email: raziz@admin.fsu.edu
Research Assistant Supervisor (if different from mentor):
Research Assistant Supervisor Email:
Faculty Collaborators: Dr. Solveig Brown She/Her
Faculty Collaborators Email: sbrown7@fsu.edu
Looking for Research Assistants: No
Number of Research Assistants: 2
Relevant Majors: Open to all majors
Project Location: On FSU Main Campus
Research Assistant Transportation Required: Remote or In-person: Partially Remote
Approximate Weekly Hours: 7, Flexible schedule (Combination of business and outside of business. TBD between student and research mentor.)
Roundtable Times and Zoom Link:
Not participating in the roundtable
Number of Research Assistants: 2
Relevant Majors: Open to all majors
Project Location: On FSU Main Campus
Research Assistant Transportation Required: Remote or In-person: Partially Remote
Approximate Weekly Hours: 7, Flexible schedule (Combination of business and outside of business. TBD between student and research mentor.)
Roundtable Times and Zoom Link:
Not participating in the roundtable
Project Description
Large Language Models are being adopted for Q&A applications in virtually every industry. Emphasis on LLM capabilities has typically focused on accuracy, safety, and aligning with human values. This project, however, will explore LLM abilities and limitations in understanding nuances in human expression. Can LLMs parse language that is poetic, sarcastic, metaphorical, or which uses slang? Can LLMs process text in a way which aligns with emotional intelligence? This project will establish a suite of tests for this kind of language to be applied to multiple large language models and compare their relative abilities.Research Tasks: Review literature on existing LLM benchmarks and LLM provider claims on what models are capable of.
Brainstorm types of nuances in human expression worth examining and develop a strategy for producing language samples.
Develop a strategy for scoring LLM responses to prompts with nuanced expression.
Skills that research assistant(s) may need: Recommended to have some experience or interest in computer programming.
Recommended ability to consider viewpoints of multiple disciplines, e.g. what nuances exist in language for sales, clinical psychology, leadership studies, and so on, and how those fields identify nuanced expressive capabilities.