The "Existential" Threat of AI Agents to Survey Research: Understanding the Problem and How to Solve It

A recent paper published in the Proceedings of the National Academy of Sciences raised a serious concern for the research industry: autonomous AI agents can now complete online surveys while passing nearly every traditional quality control. In controlled testing, these agents cleared 99.8% of attention checks, CAPTCHAs, and open-ended screening while generating responses that appeared indistinguishable from human participants.

This session looks at what that risk actually means for your business. Drawing on real behavioral data from AI agents attempting to complete surveys, you’ll see how these systems move through questionnaires and the distinctive patterns that emerge.

For teams that rely on survey data to guide decisions, the rise of AI-generated responses changes how data quality must be evaluated. This discussion will examine why traditional checks are becoming less reliable and what new signals can help identify synthetic respondents. You’ll leave with a clearer understanding of how AI-driven survey behavior appears in real research environments and what steps organizations can take to protect the integrity of their data.

What You’ll Learn

• What recent research reveals about the ability of AI agents to pass traditional survey quality checks.
• The behavioral signatures that distinguish AI agents from human respondents.
• Practical approaches for detecting and mitigating synthetic respondents in online surveys.