Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

"Girl, I'm so Serious": CARE, a Capability Framework for Reproductive Equity in Human-AI Interaction

Alice Zhong, Phoebe Chen, Anika Sharma, Kandyce Brennan, Snehalkumar 'Neil' S. Gaikwad

Abstract

Sexual and reproductive health (SRH) remains shaped by structural barriers that leave many without judgment-free information. AI chatbots offer anonymous alternatives, but access alone does not ensure equity when socioeconomic determinants shape whose capabilities these tools expand or constrain. Conventional methods for evaluating human-AI interaction were not designed to capture whether technologies holistically support reproductive autonomy. We introduce CARE, Capability Approach for Reproductive Equity, developing capabilities, functionings, and conversion factors into a Normative Design Lens and an Evaluation Lens for AI in SRH contexts. Evaluating SRH-specific non-LLM chatbots, general-use LLMs, and search engine features along credibility and reasoning, we identify two epistemic harms: source opacity and response rigidity. We conclude with design and evaluation recommendations, participatory auditing strategies, and policy implications for high-stakes domains where AI intersects with inequity.

"Girl, I'm so Serious": CARE, a Capability Framework for Reproductive Equity in Human-AI Interaction

Abstract

Sexual and reproductive health (SRH) remains shaped by structural barriers that leave many without judgment-free information. AI chatbots offer anonymous alternatives, but access alone does not ensure equity when socioeconomic determinants shape whose capabilities these tools expand or constrain. Conventional methods for evaluating human-AI interaction were not designed to capture whether technologies holistically support reproductive autonomy. We introduce CARE, Capability Approach for Reproductive Equity, developing capabilities, functionings, and conversion factors into a Normative Design Lens and an Evaluation Lens for AI in SRH contexts. Evaluating SRH-specific non-LLM chatbots, general-use LLMs, and search engine features along credibility and reasoning, we identify two epistemic harms: source opacity and response rigidity. We conclude with design and evaluation recommendations, participatory auditing strategies, and policy implications for high-stakes domains where AI intersects with inequity.
Paper Structure (16 sections, 3 figures, 4 tables)

This paper contains 16 sections, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. CARE: A Capability-Based Framework for Design and Evaluation of SRH Technologies
  4. Normative Baseline via CARE's Design Lens
  5. Evaluation Setup
  6. Tool and Question Selection
  7. Evaluation Dimensions
  8. Results
  9. Credibility
  10. Reasoning
  11. Discussion
  12. Conclusion
  13. Capability Approach Modules
  14. Additional Methods Table
  15. Citation Breakdown by Tool
  16. ...and 1 more sections

Figures (3)

  • Figure 1: CARE design and evaluation lenses. The Normative Design Lens follows the capability approach backwards, beginning with wellbeing (reproductive freedom) as the end design goal. It then identifies the functionings (outcomes actually achieved), capabilities (real freedoms to achieve desired outcomes), conversion factors (conditions that mediate that access), and resources (what users can access) necessary to reach the end design goal (see Table \ref{['tab:questionsfunctioningcapabilities']} for the capability set and example functionings). The Evaluation Lens follows the capability approach forward, beginning with resources to examine how they are mediated by conversion factors to become capabilities, which enable functionings, leading to reproductive wellbeing.
  • Figure 2: Selected credibility behaviors. Interaction (1, purple) from Layla’s Got You shows the chatbot providing no sources and implicitly asking the user to trust it without justification. Interaction (2, orange) from Claude falsely claims it cannot cite external sources. Both illustrate epistemic harm to users.
  • Figure 3: Proportion of citations from reputable sources (.org, .gov, .edu, .int) by tool. Claude reached 74% but cited inconsistently; among tools that cited without prompting, Copilot was highest at 57%.