Abstract
My technical report: AI-Smart Classroom Initiative (ASCI) Advancements, shows
improvements made to a chatbot system that increases course engagement and assists
students in office hours. The system includes a RAG chatbot trained on course content, an
office hour queue, tools for tracking student progress on Piazza and Gradescope, and a quest
system that aims to gamify course engagement. The advancements added features such as
content scraping, Discord activity monitoring, course archiving, an improved questing system,
and overall user interface upgrades. Stakeholder and user testing proved that our additions
made the system more practical and efficient for instructors, while also increasing engagement
for students. The report also includes limitations such as small testing sample sizes, and
untested long-term outcomes that leave room for future improvements.
My STS research paper examines how responsibility gets distributed and obscured in
cases of AI violence using a case study. The case is on Sewell Setzer, a fourteen year old boy
that committed suicide after developing a dangerous relationship with a chatbot developed by
Character Technologies. The paper uses both Actor Network Theory and Kantian deontological
ethics to analyze how developers, corporations, the legal system, and the user all played a role
in the case of harm that occurred. The analysis showed that Character.AI may have deliberately
engineered their chatbot to maximize emotional engagement while ignoring any safeguards to
protect vulnerable users such as Sewell. It also revealed that the current legal system is
unprepared to handle cases related to AI violence and leaves gaps in accountability. Overall, the
paper warns society on the dangers of AI and that these systems need to be designed with
responsibility.
Both papers ultimately tackle the same question: how should AI be designed
responsibly? The ASCI demonstrates one answer in practice. A chatbot system with instructor
oversight and controlled course content with the goal of guiding students to improved academic
performance. My STS paper tackles this same question from an ethical perspective. The paper
warns that responsible AI development requires integrating user safety into the design process
from the start rather than treating it as an afterthought. Together, these papers show that AI has
the potential to help people; however, that potential depends on the values and intentions of the
people who build the system. As a student studying computer science, engaging in both the
development of an AI system and analyzing its ethical drawbacks prepares me for building
systems that affect a large amount of lives.
