As AI-generated images, videos and online profiles become increasingly convincing, distinguishing real content from fabricated material is becoming more difficult. A new interactive challenge at Amsterdam’s NEMO Science Museum invites visitors to investigate this problem themselves, combining a game-based learning experience with insights into emerging AI detection techniques.
In an online gaming environment, participants are challenged to identify AI-generated characters posing as real players. By examining visual details and searching for clues, they learn how to recognise potential deepfakes and approach online photos, videos and profiles with a more critical eye.
The AI Challenge was developed in collaboration with the University of Amsterdam (UvA) and the Netherlands Forensic Institute (NFI). The initiative responds to a growing challenge in the digital information environment: generative AI can produce increasingly realistic content, making it harder for people to determine whether what they see online is authentic.
By turning deepfake detection into an interactive experience, the challenge introduces younger audiences to the technology behind manipulated media and encourages them to question the content they encounter online.
The challenge also introduces visitors to research exploring how deepfakes can be detected when visual inspection alone is insufficient. Researchers at the UvA and NFI are investigating techniques such as blood-flow detection, which analyses subtle colour changes in the skin associated with a person’s heartbeat.
These changes can provide clues about whether a video depicts a real person or has been manipulated using AI. The approach illustrates how researchers are exploring signals that may be difficult to detect with the naked eye to distinguish authentic footage from synthetic content.
By demonstrating this technique, the challenge connects public engagement with ongoing research into more advanced methods of deepfake detection.
In addition to the interactive game, visitors receive five practical tips for evaluating online content: remain sceptical, assess whether the content is plausible, look for inconsistencies, verify the source and examine the social media profile behind a post.
These steps can help people approach digital content more carefully, although they cannot guarantee that a deepfake will be identified. High-quality synthetic media can be difficult to recognise, making continued research into detection methods important.
The initiative highlights a broader challenge for AI adoption: technical advances need to be accompanied by the knowledge and critical skills that enable people to navigate an increasingly AI-mediated information environment.
The AI Challenge demo was designed for visitors aged eight and above and was on display in NEMO Science Museum’s Auditorium until 25 October 2026. By bringing AI research into an accessible museum setting, the collaboration demonstrates how scientific knowledge can help younger audiences better understand the opportunities and challenges of generative AI.
Read also
