View The Document

Accession Number:

AD1151202

Title:

Influencing Trust in Human and Artificial Intelligence Teaming Through Heuristics

Author(s):

• Thompson, Joel E

Author Organization(s):

• Naval Postgraduate School

Report Date:

2021-06-01

Abstract:

This thesis analyzes potential methods intended to influence trust within military units and their use of artificial intelligence (AI) systems. AI systems are being developed to enhance the human decision-making process and when employed properly can greatly increase the rate at which actions are taken, a key requirement for generating combat power. Human and AI teams rely on the users trust for the AI system, and that trust is influenced by rational, affective, and normative trust factors. This thesis examines those trust factors and determines that only rational trust factors are directly connected to the trustworthiness of the AI and that the users trust can be influenced independently of the AIs trustworthiness through affective and normative trust factors. Influencing the users trust of the AI through substitution of affective and normative trust factors in place of rational trust factors produces unjustified trust because this trust is not dependent on the trustworthiness of the AI.

Document Type:

Conference:

Journal:

Pages:

87

File Size:

2.19MB

Contracts:

Grants:

Descriptors:

• information science
• neural networks
• information systems
• psychology
• cognition
• artificial intelligence
• machine learning
• computer languages
• computers
• artificial intelligence software
• bayesian networks
• dimensionality reduction
• data science
• standards
• warfare
• data mining
• information processing

SubjectCategory:

• Mathematical and Computer Sciences

Communities of Interest:

• Cyber

Distribution Statement:

Approved For Public Release

View The Document