Date of Award
2024
Document Type
Thesis
Degree Name
Master of Arts (MA)
Department
Psychology
Committee Chair
Lisa Vangsness
Committee Member
Jodi Price
Committee Member
Nathan Tenhundfeld
Research Advisor
Lisa Vangsness
Subject(s)
Trust, Automation--Psychological aspects, Automation--Human factors, Natural language processing (Computer science)
Abstract
This thesis explores the measurement of trust in the context of human interactions with automated systems. Trust is influenced by factors like perceived ability, benevolence, and integrity which presents a measurement challenge. Given the lack of a universally accepted model for trust, there is an abundance of trust measurement methods in the literature. This research aimed to ascertain the reliability of NLP-based measurements to determine whether such methods can shed light on the dimensionality of trust as a construct and to isolate a Linguistic Fingerprint™ of trust. The results lent support to trust as a unidimensional construct. There was little predictive difference between congruent models (e.g., predicting trust using trust sentences) and incongruent models (e.g., predicting trust using distrust sentences). In addition to contributing to the theoretical debate as to the nature of the construct of trust, these findings have implications for the feasibility of developing real-world trust measurement tools.
Recommended Citation
Koehl, Derek, "Analyzing trust in automation using large language models" (2024). Theses. 655.
https://louis.uah.edu/uah-theses/655