Using Stanford’s research on digital twins, the partnership will expand its application to marketing, with the aim of creating and validating digital twin panels – virtual representations of survey respondents.

Ipsos and Stanford will also look to identify and mitigate the potential risks and limitations of digital twins.

The research project, built on the foundation of Ipsos’ Knowledge Panel, will initially launch in the US, followed by a global expansion.

Ben Page, chief executive at Ipsos (pictured), said the joint project “solidifies Ipsos’ leadership position in applying AI and synthetic data to market research”.

Page added that the company’s “safe and agnostic technology platforms”, data science teams and data assets puts it in “a singular position” to develop “safe, powerful, and culturally accurate digital twin panels”.

In 2024, Stanford researchers developed an agent architecture that simulated the attitudes and behaviours of over 1,000 individuals, using qualitative interviews and a large language model.

Discussing the partnership with Ipsos, Professor Robb Willer, director of PASCL and professor of sociology, psychology, and business at Stanford University, said: “We're building a future where reliable synthetic data empowers researchers and businesses to make more informed decisions, grounded in a deeper understanding of human behaviour.”

Willer added that the partnership was “committed” to responsibly addressing the risks and limitations of synthetic data technology.