Samantha D'Alonzo
Interdisciplinary PhD Student, Northeastern University
J.D. Candidate & Academic Scholar, Columbia Law School
I study systemic approaches to regulating emerging technology—drawing on AI, human-computer interaction, and law. My research examines how generative AI reshapes online discourse and the information ecosystem, and how technical and policy interventions can help people navigate it.
I am a Research Assistant in the Plural Connections Group and the AI-Media Strategies Lab. Before returning to academia, I spent three years as a Strategy & Product Analyst at Jane Street Capital, and I hold a B.S. in Mathematics from MIT.
Right now, I am focusing specifically on transparency as a regulatory goal and what scaffolding might be required, such as visualization tools for understanding text disclosure requirements, to help turn AI disclosure requirements on safety plans or incident reporting into actual accountability. I am also interested in AI alignment. Feel free to reach out with any questions or overlapping research interests!
Publications & Research
-
Helpful, Harmless, Honest? RLHF as Survey Design and Content Moderation
A systematic review of OpenAI's alignment publications, arguing that Reinforcement Learning from Human Feedback functions as a form of survey design and content moderation. Accompanied by a network-analysis visualization of the alignment literature.
Accepted · ACM FAccT 2026
-
Detecting and Enhancing Intellectual Humility in Online Political Discourse
A machine-learning classifier that identifies intellectually humble communication patterns in text, contributing to research on improving the quality of online discourse. Led a team of nine researchers.
Read in ICWSM Proceedings → IC2S2 2025 ICWSM 2026
-
"Silicon Sampling": Guidance for Communications Practitioners Using LLMs as Human Surrogates
A white paper synthesizing 50+ technical papers into practical guidance for communications practitioners considering large language models as stand-ins for human respondents in survey research.
-
Machine-Learning Media Bias
An algorithm for automatically detecting differentially-used phrases across a corpus of over one million news articles, using information scores and novel NLP techniques to quantify media bias.
-
Simulating and Evaluating Rebalancing Strategies for Dockless Bike-Sharing Systems
Peer-reviewed research presented at the Transportation Research Board Annual Meeting on optimizing the placement and rebalancing of dockless bike-sharing systems.
In Progress
-
Paradigm Shifts in the AI-Media Era
A manuscript in preparation for Social Media + Society exploring how the proliferation of generative-AI content is driving an epistemic shift in the media industry, and how technical solutions such as provenance and watermarking can help users navigate AI-generated content.
Target: Social Media + Society
Invited Talks
-
Helpful, Harmless, Honest? RLHF as Survey Design and Content Moderation
-
From Training to Model Cards: Design Choices, Content Moderation, and Accountability in LLMs
News
- Aug 2026Beginning my J.D. at Columbia Law School as one of two students selected for the Academic Scholars Program.
- May 2026Gave a talk, "Helpful, Harmless, Honest? RLHF as Survey Design and Content Moderation," for Professor Booth's Giraffe Lab at Brown University.
- Apr 2026Gave a talk, "From Training to Model Cards: Design Choices, Content Moderation, and Accountability in LLMs," for Professor Wang's group at Cornell Tech.
- Mar 2026Presented "Detecting and Enhancing Intellectual Humility in Online Political Discourse" at the Intellectual Humility Conference in Palm Springs.
- 2026Paper on RLHF as survey design and content moderation accepted to ACM FAccT 2026.
- 2026Intellectual humility paper accepted to ICWSM 2026.
- 2025Presented the intellectual humility work at IC2S2 2025.
- 2024Joined Northeastern University as an interdisciplinary PhD student in the Plural Connections Group.
Writing
On my Substack I write accessible primers on AI policy and technology issues for a general audience—translating technical research into plain language.