Everyone has an opinion on artificial intelligence, or AI, but for researchers who study AI policy it’s not clear where those opinions begin.
Tina Law, an assistant professor of sociology at UC Davis, was working on a paper about AI governance when she realized that it’s not clear where people get their information about the technology. Those sources of information could limit oversight by the general public — people outside the development and scaling of AI.
“Publicly available content on AI is largely created by tech companies that develop AI and that have a vested interest in policies that might regulate their businesses,” said Law. “It's a conflict of interest if they’re not providing objective information that could inform the average American on what the harms or benefits of AI are.”
With support from a new University of California Office of the President Early Career Faculty Research Excellence Award, Law expands her research on AI governance with a new research project in which she is constructing and analyzing a database of more than 50,000 YouTube videos to learn how they contribute to the general public’s understanding of AI.
Learning what YouTube videos explain about AI
YouTube videos are at the center of this project because the platform has become a primary source of information for people seeking to learn about new topics and to build new skills. As of October 2025, YouTube was the fourth-largest social media platform with roughly 2.6 billion monthly active users. A recent report on social media from the Pew Research Center found that nearly half of U.S. adults use the site daily.
This study seeks to answer four research questions:
- What information about AI is available on YouTube?
- Who are viewed as AI “experts”?
- How is AI explained?
- How do users interact with AI experts?
This project will use manual coding and AI models to analyze data from the videos, including video frames, transcripts and viewer comments from more than 50,000 YouTube videos about AI. Law worked with UC Davis DataLab senior data scientist Nick Ulle to build this database of videos posted between 2019 and 2025, right around the public release of ChatGPT and other AI tools.
Based on Law’s preliminary analysis, the content of the videos is highly diverse. Some are educational, such as explainers on machine learning and AI, while others share how to start a business using AI. There are also some videos explaining AI that are synthetic or completely created by AI.
Some of the videos are produced by the companies developing AI technology, such as Google. In these videos, said Law, explaining AI can be a way for companies to build business for their own products. However, Law said there may be a conflict of interest if those explainers influence whether the general public sees a need for AI regulation.
Funding from the Early Career Faculty Research Excellence Award will support Law’s work analyzing this new database. The funding will also make it possible to hire research assistants and fund computing resources and technical support from the UC Davis DataLab.
Law is currently working with undergraduate research assistant Ela Wang to manually label a small sample of videos that will help in automatically classifying the content of all videos in the database.
Increasing participation in how AI affects everyone
Right now, said Law, there are no federal laws that require public oversight and regulation of AI as it’s developed by current major players in the field, such as OpenAI, Anthropic, Meta and Google. These are also companies that have invested substantially in lobbying government. A Washington Post report found that in 2021, leading tech firms reportedly spent nearly $70 million combined on lobbying.
In addition to helping build a foundation for broader participation in AI oversight, the project will also expand what we know about how social media platforms affect public understanding of new technologies more broadly.
“Everyday Americans should have say in what happens with AI because it's affecting so many parts of their lives,” said Law. “The first thing we need to do as social scientists is to learn how people come to develop an understanding of what AI is.”
YOU MAY ALSO LIKE THESE STORIES
Does Structural Racism Shorten People's Life Expectancy?
While structural racism can be measured across an entire neighborhood, city or county, its impact on people’s lives might be more localized. A recent study in sociology found that people most impacted by structural racism died more than two years sooner than the average American adult.
The Untold Story of How COVID-19 and Disinformation Nearly Undermined Taiwan’s Democracy
The COVID-19 pandemic was a global crisis that showed the fragility of democracy in Taiwan, where misinformation threatened the three pillars that define the nation’s government and civil society.