Seed Grants

Jodie Righos / Unsplash

Seed Grants

To support independent scholarship on how information, technology and democratic practice interact, IIID awards seed grants to our Northeastern affiliate network.

Seed projects include a diverse range of research topics and disciplines with a strong emphasis on public policy relevance; creating tools, data sets, and frameworks around the study of online behavior and democracy-related issues; and, leveraging empirical data collected through the National Internet Observatory.

2026

The Productivity-Sycophancy Trade-off: Empirical Regulatory Frameworks for AI-Mediated Work. Professor Malihe Alikhani (Computer Science) investigates whether AI tools that mirror users' biases create measurable declines in decision-making quality — and is developing a regulatory framework to help policymakers address the risks of automated bias reinforcement in professional settings.

Mapping Elite AI Discourse Across Online Political Communication. Professor Beatrice Magistro (Political Science, Public Policy and Urban Affairs) is piloting a dataset of elite AI discourse drawn from political podcasts and YouTube channels run by governmental bodies or public affairs groups, tracking how AI's effects on jobs and the economy is framed in direct-to-voter venues.

Simulating Belief Networks: New Data, Models, and Audiences. Professors Brian Ball and David Freeborn, and doctoral student Federica Imbriale (Philosophy, Northeastern University London) are convening an international workshop to advance the PolyGraphs belief-network simulation program toward integration with empirical data from the National Internet Observatory, producing a research roadmap and public-facing information literacy materials.

When Do User Corrections Work? Controlled Evaluation of Corrections Efficacy in Realistic Social Media Settings. Postdoc Lucy Butler and Professor Briony Swire-Thompson (Political Science, Psychology, Network Science Institute) use a social media simulation environment to test whether standard experimental designs overestimate the effectiveness of user corrections at reducing the spread of misinformation.

Algorithmic Literacy Across Cultures: A Three-Country Comparative Study. Professor Myo Chung (Journalism) is extending prior IIID-funded research into a three-country comparative study of algorithmic literacy, adding a nationally representative survey in Nigeria to existing U.S. and South Korean datasets.

Data Extraction in Digital Healthcare Applications: Exploring Alternatives to Legally (Un)Regulated Bargaining in the Platform Economy. Professors Hilary Robinson (Law, Sociology) and Nadir Hamid (Law) examine structural power asymmetries in platform data agreements, using a blockchain-secured health-tracking device as a case study to develop legal and technical frameworks that would give users meaningful control over their own data.

The Politics of Privacy Evaluation: How AI Researchers Define and Measure Data Privacy. Professors Jayshree Sarathy (Computer Science) and Chuncheng Liu (Communication, Sociology) examine how privacy evaluation methods gain legitimacy in AI research communities and become embedded in regulatory and governance infrastructure.

Mapping Brand Polarization and Political Advertising Effectiveness. Professor Yakov Bart (Business) is developing a sociospatial framework analyzing how brands' political valence shapes the commercial media environments in which political advertising operates, building reusable data infrastructure from Northeastern's licensed L2 consumer and voter datasets.

2025

Decoding Algorithms: Bridging the Algorithmic Literacy Gap for a Fairer Digital Landscape. Principal Investigators. Professor Myojung Chung (Journalism) investigates algorithmic literacy — understanding how algorithms work and their effects — to safeguard user autonomy, foster informed public dialogue, and uphold democratic values. This project aims to address these challenges and foster a more informed and inclusive digital landscape by building on findings from a previous study, which examined within- and between-country algorithmic knowledge gaps across four nations.

Information Exposure and Information-Seeking Behavior in User Interactions with Large Language Models. Postdoc Pranav Goel (Network Science Institute) explores information exposure and information-seeking behavior in the context of LLMs to understand types of external websites users are directed to by LLMs; whether LLMs encourage or discourage users to go to external websites, and whether there are any systematic differences between the external sources that different groups of users (based on age, gender, partisanship, or education) are exposed to in their interactions with LLMs.

Network Analysis of the AI Ethics Field. Professors David Freeborn, Brian Ball, Alice Helliwell (Philosophy, Northeastern University London), Alex Cline (Computer Science), and Research Assistant Kevin Loi-Heng (Northeastern University London) chart the landscape of the AI ethics field through automated computational methods, integrating natural language processing and network analysis to examine topical structures and map the institutional, co-authorship, and citation networks of the field. The project seeks to build (1) an open-source Python-based toolkit for analyzing the field of AI ethics (2) an open-access dataset mapping landscape of the field, (3) interactive visualizations that make complex network relationships accessible to researchers and practitioners and (4) an automated literature review of the state of the art in AI ethics research. These tools and resources will enable scholars to identify emerging ethical frameworks, uncover gaps in current research, and foster more effective collaboration across disciplinary boundaries.

SMS-to-Web Public Opinion Polling for Non-Profit News Organizations. Professors Katherine Haenschen (Communication Studies, Political Science) and Justin de Benedictis-Kessner, (Harvard Kennedy School) seek to develop a low-cost tool for non-profit news organizations to use to generate scientifically rigorous public opinion data in their communities, which they can use in their reported coverage. The tool aims to support local, regional, and statewide non-profit news organizations that offer a bulwark against the continued shuttering of reporting outlets in an era of budgetary constraints and private equity acquisition.

Semantic Search for Automated Misinformation Classification at Scale in the Wild. Jason Radford (Political Science) and Stefan Wojick (Independent Consultant) examine online misinformation by developing semantic search models tailored to misinformation claims from four key domains: hate, health, politics, and climate. These models will be evaluated across data from three major social platforms — Twitter, Reddit, and Telegram, taken during the same time period 2019 to provide an apples-to-apples comparison of the prevalence of misinformation across platforms with different user bases and affordances.

Integrating Human Gaze Patterns and Machine Learning for Enhanced Detection of Visual Bias in Online Political Advertising. Professors Sunny Yang (Political Science, Communication) and Yakov Bart (Business) address the challenge of identifying visual bias in online political advertising by combining human semantic knowledge and gaze behavior with existing computer vision frameworks. Using a subset of political ads from the Facebook Ad Library and eye-tracking experiments, the project examines how people perceive visual bias in advertisements by analyzing gaze patterns, such as fixation duration and transitions between image segments. By combining human perception data with machine learning, the project enhances automated bias detection, offering insights into visual persuasion techniques and improving media literacy to help the public critically evaluate political content.

Watchdog - AI Visualization Tools for Journalists. Professors Viraj Upadhyay and Saiph Savage (Computer Science) seek to build out Watchdog, an AI-powered visualization tool, developed to combat targeted political harassment and abuse of power, with a specific focus on protecting journalists, particularly in regions like Mexico where press freedom is under threat. Its primary function is to track mentions of journalists by political figures, analyze the sentiment, and understand the context of these mentions. By using Natural Language Processing (NLP), Watchdog can detect patterns in defamation, misinformation, narrative manipulation, and propaganda.

2024

AI as a Team Member: Human-AI Group Discussion Improves Misinformation Detection. Professor Chenyan Jia (Journalism ) examines whether 1) group discussion can improve individuals’ ability to identify false information from truth; 2) the group dynamics will change or not after adding AI as a new actor. She will build on the ChatGPT API with the function calling feature. The study intends to investigate whether collaboration can help identify misinformation and reorients research and policy from focusing on the individual to a more collaborative and social approach in addressing the problem of misinformation in the era of generative AI.

Accessing ChatGPT’s Ability to Mimic Humans on Social Media. Postdoc Kaicheng Yang, (Network Science Institute) tests the ability of ChatGPT to mimic individual Twitter users. Due to the restrictions on data accessibility from platforms like Twitter and Reddit, researchers have limited ability to study online behavior. Therefore, demonstrating the viability of using LLMs to simulate humans could facilitate cost-effective and ethically responsible experiments. LLMs might also act as an alternative source of social media data. From the trust and security perspective, the risk of adversarial actors employing LLMs to fabricate fake personas on social media is imminent. This study could yield valuable insights into how LLMs facilitate such abuses.

Vision Beyond Sight: Designing Human-Centered AI Systems for Social Media Accessibility for the Visually Impaired, through the National Internet Observatory. Professors Saiph Savage, (Computer Science) and Yakov Bart (Business) leveraging human-centered AI systems to enhance social media accessibility for visually impaired individuals. This initiative, operating under the auspices of the National Internet Observatory (especially by using visual data from the observatory), aims to bridge the gap between the rapidly evolving digital landscape and the needs of those with visual impairments. At the core of this project is the development of AI-driven tools tailored to interpret, translate, and present social media content in formats accessible to visually impaired users.

Publics and Their Opinions: Measures on Realistic Simulations. Professor Brian Ball (Philosophy, Northeastern University London) examines the (collective) opinions of various social groups, or ‘publics’, under realistic informational conditions, and to assess various measures of those aggregate attitudes. Building on the techniques and findings of the PolyGraphs project, the research uses computational methods to simulate communities of agents in their quest for informed opinions in adverse circumstances – e.g. confronted with mis- and disinformation, agents may be uncertain which of their network neighbors are trustworthy, or have opinions/expertise that should be deferred to.