Supplementary Materials

Griefbots and the Governance of Posthumous AI: A Normative Framework (2026)

This repository contains the data and code accompanying the paper "Griefbots and the Governance of Posthumous AI: A Normative Framework." The analysis explores public perceptions and emotional responses to AI-mediated interactions with digital representations of the deceased, based on a dataset of YouTube comments.

Overview

We analyzed a total of 3,574 comments from the YouTube documentaries:

• ENDRV: "Living Forever Through AI: Digital Immortality and the Future of Death (2023)", Access date: 2025-11-11
• GUARDIAN: "Back from the dead: could AI end grief? (2024)", Access date: 2025-11-12
• ABC News: "Generative AI helps grieving family remember their daughter (2024)", Access date: 2025-11-12

Our analysis combined the following:

• Sentiment Analysis using fine-tuned LLMs (RoBERTa)
• Unsupervised Topic Modeling with BERTopic

Explore the Full Analysis

First, install all dependencies by running:

pip install -r requirements.txt

To perform the full data processing and analysis, run pipeline.ipynb in Jupyter Notebook or Jupyter Lab.

To download and analyze comments from your own YouTube video, change the YouTube URL in the notebook as shown below:

URL = "https://www.youtube.com/watch?v=5udOx8-QxtE" # Change this to your desired video URL

This will automatically retrieve comments from the specified video and integrate them into the analysis pipeline.

To visualize the analysis results, run the visualization.ipynb.

Repository Contents

File	Description
data/	The dataset contains the downloaded YouTube comments, and the results of the analysis.
src/	Python scripts for data processing and analysis:
crawling.py	Script for scraping YouTube comments.
preprocess.py	Script for cleaning and preparing text data for analysis.
sentiment-analysis.py	Script for performing sentiment analysis using fine-tuned LLMs.
bertopic-analysis.py	Script for topic modeling with BERTopic.
pipeline.ipynb	Jupyter notebook demonstrating the full analysis pipeline.
visualization.ipynb	Jupyter notebook for visualizing the analysis results.
requirements.txt	List of required Python packages for running the analysis.

License

This repository is released under the MIT License.