Improvisation and trust in human/autonomy teams: a task-based perspective
Loading...
Authors
Gonzalez, Daniel, Alejandro
Issue Date
2024-08
Type
Electronic thesis
Thesis
Thesis
Language
en_US
Keywords
Industrial and management engineering
Alternative Title
Abstract
This dissertation explores and investigates the interplay between processes of improvisation and trust within collectives composed of human and synthetic agents. Observations and insights from various settings, particularly but not exclusively in the arts, suggest a close relationship between improvisation and trust. On the one hand, improvisation entails creativity (the production of novel actions) under time constraint. On the other, trust can be both an enabler and a by-product of improvisation, as–through shared experiences–team members understand each other’s capabilities, even under highly non-routine conditions. With the potential introduction of artificially intelligent (AI) (or synthetic) team members to these types of task, questions arise surrounding the conditions under which human/AI team members can (or should) improvise and build trust. Prior research on improvisation in human groups outside the domain of music has overwhelmingly been informed by opportunistic observation in the field. In contrast, this dissertation begins to pry apart the complex processes of improvisation and trust through a combined program of theory-building and human-in-the-loop experimentation. The central question this work seeks to address is “How do task-related constraints and goals determine the extent to which the behavior of human and human-AI collectives is planned or improvised and affect trust?” This research first establishes the theoretical basis that associates task constraints with the response to these constraints in human collectives (which for simplicity we refer to as groups). To do this an analysis and synthesis of prior research is performed to yield theoretical models that suggest how task-related factors —time constraint, locus of control,and workflow— shape how behavioral responses range from planned to improvised in Chapter 3. The second part of the research further expands these theoretical models with the inclusion of trust, represented with three aspects (competence, predictability and integrity), among members of human-AI groups in Chapter 4. The previous parts of the research yield a task-centric analysis of prior literature on improvisation and related areas, while also providing a set of research questions and propositions for further testing in controlled settings. The third and last part of the research presents the design and development of an experiment in order to test propositions regarding the processes of human improvisation and trust in human/AI groups and changes in the task environment. The study and results are presented in Chapter 5. Conclusions and proposed work of the overall thesis are presented in Chapter 6. This work has three contributions. The first contribution is the development of a theoretical basis to associate the design of task-related constraints with (improvised) behavioral responses by human or human-AI groups. The second contribution is a parallel exploration of the aspects of trust and the behavioral measurements developed for these aspects in a human-AI group setting. The third contribution is the development of an experimental paradigm that allowed the evaluation of behavioral measurements of the aspects of trust for a human-AI team in a creative task.
Description
August 2024
School of Engineering
School of Engineering
Full Citation
Publisher
Rensselaer Polytechnic Institute, Troy, NY