Abstract
Computational systems do not simply produce information; they shape how people understand problems, make decisions, and assign responsibility. My capstone project and my Science, Technology, and Society (STS) research paper both examine this idea from different perspectives. My capstone project is a NASA-focused tutorial that uses Harmonized Landsat-Sentinel 2 (HLS) data in Google Earth Engine to analyze vegetation over time. I undertook this project to better understand how satellite imagery and cloud-based tools can support environmental monitoring in a clear and reproducible way. My STS research paper examines how responsibility for AI harm is shared among governments, corporations, and users. I pursued this research to understand how accountability affects real-world harm, especially for emotionally vulnerable users. These projects are connected because both focus on how computational outputs are processed, interpreted, and trusted.
My capstone project addresses the challenge of turning large amounts of satellite imagery into a usable analysis process. It provides a structured tutorial for accessing and processing HLS data using Google Earth Engine and Python. The workflow includes defining a region of interest, filtering imagery, applying cloud masking, and calculating vegetation indices such as NDVI and EVI. It then converts the data into a time series for analysis. Because satellite data often has gaps, the workflow uses resampling and interpolation to create a more continuous dataset. This allows users to observe vegetation patterns more clearly over time.
The main conclusion of my capstone project is that HLS data can support effective vegetation monitoring when processed through a structured workflow. By combining Landsat and Sentinel imagery, HLS improves observation frequency and data coverage. The project also shows that reliable results depend on proper preprocessing and careful handling of missing data. Using time-series methods, the workflow identifies seasonal patterns, long-term trends, and anomalies. Overall, the project demonstrates that remote sensing tools can provide meaningful environmental insights when used carefully and consistently.
My STS research paper asks how governments, corporations, and users assign responsibility for preventing harm in AI systems, and how this affects emotionally vulnerable users. This question is important because AI systems are widely used and can influence decisions and well-being. The research uses qualitative document analysis of policy documents, corporate agreements, academic studies, and real-world cases. It is guided by a risk governance framework, which argues that responsibility should align with power and control.
The findings show that responsibility for AI harm is unevenly distributed. Governments recognize AI risks but often lack strong enforcement, while corporations control system design but limit their liability. As a result, users are left to manage risks they may not fully understand. This is especially serious for emotionally vulnerable users who may rely on AI systems for support. The research concludes that AI harm is not only a technical issue but also a structural one, and that stronger institutional accountability is needed to better protect users.
