How good are humans at detecting AI-generated images? Learnings from an experiment

TL;DR

This study assesses human ability to distinguish AI-generated images from real ones through a large-scale online experiment, revealing modest success rates and highlighting challenges in detecting AI images, especially landscapes.

Contribution

It provides empirical data on human detection accuracy of AI images at scale, emphasizing the difficulty and need for transparency tools in AI-generated visual content.

Findings

Overall detection success rate of 62%

Highest accuracy with human portraits

Significant difficulty with landscapes

Abstract

As AI-powered image generation improves, a key question is how well human beings can differentiate between "real" and AI-generated or modified images. Using data collected from the online game "Real or Not Quiz.", this study investigates how effectively people can distinguish AI-generated images from real ones. Participants viewed a randomized set of real and AI-generated images, aiming to identify their authenticity. Analysis of approximately 287,000 image evaluations by over 12,500 global participants revealed an overall success rate of only 62\%, indicating a modest ability, slightly above chance. Participants were most accurate with human portraits but struggled significantly with natural and urban landscapes. These results highlight the inherent challenge humans face in distinguishing AI-generated visual content, particularly images without obvious artifacts or stylistic cues. This study stresses the need for transparency tools, such as watermarks and robust AI detection tools to mitigate the risks of misinformation arising from AI-generated content