Graduation Year

2026

Date of Submission

4-2026

Document Type

Campus Only Senior Thesis

Degree Name

Bachelor of Arts

Department

Mathematics

Second Department

Computer Science

Reader 1

Mark Huber

Terms of Use & License Information

Terms of Use for work posted in Scholarship@Claremont.

Rights Information

© 2026 Myrvens Sylvestre

Abstract

This study analyzes the challenges that come up when companies use artificial intelligence agents, systems that can make decisions and take actions with limited human direction, to support and carry out important tasks. Unlike earlier AI systems that only helped humans, AI agents can now act on their own, work across different systems, and make decisions continuously at a large scale. While this can improve efficiency and performance, it also creates new risks related to accountability, transparency, bias, and reduced human oversight. This thesis argues that current AI ethics frameworks are not well suited for these systems because they treat ethics as rules applied after deployment instead of building them into the system from the start. The study identifies key challenges, including risks to privacy and data protection, biased outcomes, limited explainability, unclear responsibility, and over-reliance on automated decisions. Bias is not only a data issue but also a broader problem shaped by historical patterns, system design, and feedback loops that can grow over time if nothing is done to reduce it as much as possible. Using conceptual analysis and two enterprise case studies, Oracle Fusion Cloud and a multi-agent AI system, this research shows how the autonomy and coordination of AI agents make these challenges more serious in real-world settings. In response, this thesis proposes a solid design framework based on bounded autonomy, explainability and auditability, controlled tool access, and ethics by design. It also highlights the need for continuous monitoring and clear governance structures to maintain responsibility while keeping humans involved in decision-making. Overall, the findings show that strong system design, oversight, and accountability are necessary to ensure AI agents operate in reliable, transparent, and socially responsible ways.

Download

This thesis is restricted to the Claremont Colleges current faculty, students, and staff.

Share

COinS