Approved Projects

This project investigates user behavior on and around TikTok during its temporary U.S. ban, examining shifts in platform usage and changes in content creation and engagement before, during, and after the event

This study explores Americans’ news consumption on social media, analyzing the roles of traditional versus influencer media, and how platform use correlates with demographic and political variables

This project examines how toxic versus non-toxic online behaviors correlate with user personality traits and engagement patterns, aiming to identify ways to design more constructive digital interactions

This study investigates how general-purpose AI systems like ChatGPT are actually used in the real world, examining temporal, geographic, and thematic trends in usage to better align AI development and safety evaluations with real-world needs

This study evaluates Google’s Topics API by comparing its claimed privacy protections with real-world user data, aiming to identify potential privacy risks and assess the uniqueness and stability of user profiles over time

The project investigates how users seek news through AI-driven tools like ChatGPT and Google AI Overviews, aiming to understand media linking, user interaction, and demographic trends in AI-mediated information access

This study assesses how exposure to online content – categorized by credibility, political leaning, geography, and AI-generation – varies across demographic groups, with the goal of understanding disparities in information consumption.

This project analyzes how the rise of large language models (LLMs) like ChatGPT is influencing user behavior around traditional search engines, focusing on search category shifts and demographic adoption patterns.

The study examines behavioral changes following exposure to AI-generated content in search results and chatbots, including patterns of information consumption and their broader societal implications

This study quantifies the extent of echo chambers and demographic segregation in digital information environments by analyzing what users are exposed to, rather than what they engage with. It examines how exposure varies across age, gender, and content type to reveal structural patterns that have remained obscured by earlier research focused only on digital trace data.

This project aims to understand the impact of LLM-based chatbots on the open web, especially from the lens of users’ propensity to explore external websites.

This project provides a descriptive account of desktop computer usage using NIO panel data, documenting how people spend time across websites and demographic groups.

We study how demographic variables can inform online attention across information-rich vs. hedonic categories and how this attention evolves and reacts to shocks. This research intends to shift the digital-inequality conversation from “who is online” to how scarce attention is allocated over time, yielding actionable levers for inclusive marketing and policy.

We explore how generative AI technologies are reshaping the ways in which people seek information, and how this, in turn, influences our understanding of the world around us. Our goal is to provide policy insights that can help guide the use of AI tools in the service of humanity’s pursuit of what is true, worthy, and beautiful.

People often feel worse after reading the news, yet they still keep scrolling. This project seeks to understand why. Does exposure to negative news trap people in a cycle of bad mood and more negative reading (“doomscrolling”), or do they instead pull back to protect their emotional well-being? Using large-scale online browsing data, this study will track how people’s moods and news choices influence each other over time.

This project examines how AI-written summaries at the top of search engine result pages affect how consumers search for and read customer reviews. The goal is to see whether these summaries shape information gathering and shopping decisions.

We study how people’s online habits change after they start using AI tools like ChatGPT, including what kinds of news and websites they choose to visit and how this affects the quality and balance of information they consume and are exposed to. Using anonymous, pseudonymized browsing data, we look at whether AI use is linked to more or less political polarization, more or less consumption of trustworthy sites, and changes in how much people rely on social media and online communities, and how these patterns differ across groups of people.

This project analyzes how gen-AI search experiences reshape consumers’ online journeys from initial exploration to product evaluation and how these changes affect downstream engagement with merchants, reviews, and paid versus organic landings. The study leverages NIO time-on-page and URL trails to estimate effects around documented GenAI rollouts and retailer feature launches.

We look at what people read on news websites and for how long to learn whether groups are seeing the same news or very different stories. This can help explain why people often have different views of the same events and how we might reduce that divide.

This project aims to study the quality of search results, with a focus on the AI summaries that increasingly appear on Google and Bing, and a topical focus on health-related searches.

This study looks at how Americans get political news and see campaign ads during recent elections. Using web-browsing data and detailed records of political advertising from 2018–2024, we study what news people read, which ads they are likely to encounter, and how today’s political media environment differs from the past.

KIND-AI studies how different people use AI chatbots by looking at their conversations, online browsing, and survey responses. We want to understand who uses these tools, what they talk about, their mental health and personality profiles, and how their interactions change over time.

We’re studying how often people use AI chatbots (like ChatGPT and similar tools) as a substitute for web search. By estimating what share of chatbot questions are really “search” and comparing that to Google use, we’ll help show how much competition AI creates in online search.

We seek to understand how the types of questions posed to online search tools have evolved with increasing usage of AI chatbots and answer engines to augment (replace?) traditional search engine utilization.

This project examines how people search for reproductive health information online and what content major search engines and large language models provide in response. Using data from the National Internet Observatory, we map the informational pathways that shape public understanding and access to reproductive care.

This study examines whether AI chatbots like ChatGPT may unintentionally narrow how people explore questions about their own identity – including their political views, religion, or mental health – and whether this affects how they later search for information on Google.

This project studies what news people actually read and watch across the internet–not just what they post on social media–and how big political events and AI chatbots shape those habits. The goal is to see whether “filter bubbles” hold up in private browsing and whether chatbots push people toward, or away from, balanced, high-quality information.

This study examines how much effort people must use when interacting with AI tools such as ChatGPT and Gemini. We analyze conversation patterns to understand whether some groups face greater difficulty getting useful help from AI systems and whether that burden relates to well-being.

This project aims to study how AI companies compete in the marketplace and its implications for consumers.

This project examines how young adults learn about emotions and relationships through AI chatbots. It also explores how differences in access to social support, across gender, race, and time of day, may shape these experiences.

This study looks at how chatbots like ChatGPT and Gemini answer questions differently from Google and Bing when people ask about topics where experts disagree. We also look at whether people have started searching differently, in longer and more conversational ways, since chatbots became popular.

This study looks at what people search for online when it comes to Jews, Judaism, and Israel, and what information search engines show them. We want to understand whether people are getting accurate and balanced information on these topics.

How do people use AI to learn about the world? They ways they do tell us about our culture, as well as how AI may ultimately influence people’s beliefs and interactions.

LLMs are social actors, and users often have real relationships with them. However, they are also pieces of software; this study examines how people’s interactions with chatbots change after a model of software update disrupts the chatbots memory and personality.

This study examines whether people who are exposed to AI overview on search engines like tend to look up or fact-check the information those tools give them and adopt a causal design to assess latent features in the overview that affect original source reading behavior.

The project investigates how GenAI chatbots such as ChatGPT and Google Gemini act as epistemic agents - that is, how they shape what counts as credible knowledge in interaction with users. It aims to understand the changing dynamics of authority and trust as people increasingly turn to AI systems for information and interpretation.

When big political crises happen, people are increasingly asking AI chatbots for help to understand the news and the world around them. Our study looks at how people talk to chatbots like ChatGPT during emergencies to make sense of complex national events, and whether these conversations show signs of national anxiety.

This project studies how different social and minority groups—defined by religion, migration background, social class, and gender—use the internet, Google Search, and AI tools like ChatGPT and Google Gemini in everyday life. The goal is to understand who relies on which tools and how this may create new inequalities, without identifying individual people.

As AI chatbots become involved in more people’s lives through virtually all avenues, the effect on people’s lives remains to be fully detailed. This study aims at understanding how people use chatbots, how those usages change over time, and what impacts this may have on users.