SearcharxivSearch

arXiv subjects

Megan Levis Scheirer

Publications and source records attributed to Megan Levis Scheirer.

2 recordsLinked to original sources

Do We Need Subsidiarity in Software?

Subsidiarity is a principle of social organization that promotes human dignity and resists over-centralization by balancing personal autonomy with intervention from higher authorities only when necessary. Thus it is a relevant, but not previously explored, critical lens for discerning the tradeoffs between complete user control of software and surrendering control to "big tech" for convenience, as is common in surveillance capitalism. Our study explores data privacy through the lens of subsidiarity: we employ a multi-method approach of data flow monitoring and user interviews to determine the level of control different everyday technologies currently operate at, and the level of control everyday computer users think is necessary. We found that chat platforms like Slack and Discord violate subsidiarity the most. Our work provides insight into when users are willing to surrender privacy for convenience and demonstrates how subsidiarity can inform designs that promote human flourishing.

cs.HC

Is It Possible to Make Chatbots Virtuous? Investigating a Virtue-Based Design Methodology Applied to LLMs

With the rapid growth of Large Language Models (LLMs), criticism of their societal impact has also grown. Work in Responsible AI (RAI) has focused on the development of AI systems aimed at reducing harm. Responding to RAI's criticisms and the need to bring the wisdom traditions into HCI, we apply Conwill et al.'s Virtue-Guided Technology Design method to LLMs. We cataloged new ethical design patterns for LLMs and evaluated them through interviews with technologists. Participants valued that the patterns provided more accuracy and robustness, better safety, new research opportunities, increased access and control, and reduced waste. Their concerns were that the patterns could be vulnerable to jailbreaking, were generalizing models too widely, and had potential implementation issues. Overall, participants reacted positively while also acknowledging the tradeoffs involved in ethical LLM design.

cs.HC