Comment reconnaitre une vidéo créée par IA ? (j'ai le truc !)

Comment reconnaitre une vidéo créée par IA ? (j'ai le truc !)

by Ludo IA²

Humanities, Social Sciences & Thought AI — Models & Research UYQArtificial intelligence (AI)UYQVComputer vision
🎙 Ludo IA² 👥 8K 📅 January 23, 2026 ⏱ 10 min 👁 3K 📄 expert opinion 🧭 2026-08-21
Available in: English (current) Français

Keywords

AI videodetectionRunwayTuring Reeldeepfake

Summary

In this video, the creator, Ludo, presents a live test from Runway’s ‘Turing Reel’ to identify AI-generated videos. He plays several rounds, guessing which of two videos is AI-made, and shares his reasoning based on visual cues like lighting, motion, and details. He scores 65% correct, admitting the difficulty. He concludes that AI videos often have a ’too perfect’ quality and encourages viewers to take the test and share results. He emphasizes the importance of educating people to recognize AI content on social media to prevent manipulation. The video is informal and aimed at raising awareness rather than providing a rigorous scientific analysis.

103 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video’s value lies in its practical demonstration of a real tool and its potential to raise awareness about AI-generated content. However, the argumentation is weak: the creator’s ’trick’ is based on subjective impressions and a small sample, not on systematic analysis. He even admits to being wrong multiple times, undermining the reliability of his method. The reasoning is often speculative (e.g., ’too perfect’ lighting) and lacks empirical support.

Scientific Rigor, Source Quality, Title Accuracy

The video cites Runway’s Turing Reel as the test source, which is a legitimate reference. However, no other scientific sources are mentioned. The title promises a ’trick’ but the content delivers only anecdotal observations. The video’s scientific rigor is low; it is more of a personal experience than a structured study. The description includes links to other videos and resources, but these are not cited in the video itself.

153 words

Title / Content Match

The title accurately reflects the content: the creator demonstrates a method to spot AI-generated videos, though the 'trick' is subjective and not scientifically validated.

Quality & Reliability

5/10

The video is a personal, anecdotal test of an AI detection tool, with no scientific methodology or verifiable data. The creator's conclusions are based on subjective visual cues and a small sample, lacking rigor.

Chapters

Cited Sources

Concurring Sources

  • Runway Turing Reel — The test is a real tool from Runway, supporting the video's premise.

Contribution & Novelties

The video offers a personal, live demonstration of an AI detection test, which is engaging but not novel. It highlights the difficulty of distinguishing AI-generated videos, a growing concern. The ’trick’ proposed (looking for ’too perfect’ lighting) is not new and lacks validation.

Pour aller plus loin :

  • Deepfake detection — Overview of deepfake technology and detection methods.
  • Media literacy — Skills to critically evaluate media content, relevant to the video’s call for education.
  • Runway Research — Official research page, including the Turing Reel test.

85 words

Radar Profile

The radar profile shows low scores across all dimensions, with slightly higher for information quantity and technical level, reflecting the video's informal and anecdotal nature. The low reliability score indicates a lack of scientific rigor.

Reliability 3/10