How Pew Research Center is – and is not – using AI in our work

Pew Research Center describes a deliberate, people-centered approach to artificial intelligence: surveys are answered by real people (not synthetic models), U.S. polls rely on a multimode, probability-based panel of about 10,000 adults, and humans decide research topics, design analysis plans, and write and review reports. The organization emphasizes accuracy, methodological rigor, respondent privacy (personal identifying information is never exposed to public AI tools), and the use of real photographs and illustrations rather than AI-generated imagery. Responsibility for quality and interpretation remains with staff researchers and editors.

In practice, AI is already used in limited, controlled ways. Engineers use AI code assistants to develop the website; researchers may use AI to help write code for preparing and analyzing datasets and to assist with textual tasks such as coding open-ended responses and scraping web content, while humans retain design and interpretive control. AI tools aid grammar and initial copy editing and help adapt human-authored findings into derivative products like social posts, but they do not produce original analysis and all AI-assisted content is reviewed by people. The Center commits to disclosing meaningful AI use in methodology sections and to updating its public guidance if AI use expands materially, and it invites public feedback.