Select Language:
Generative AI is advancing rapidly, making it increasingly hard for many of us to tell whether we’re interacting with a human or an AI chatbot. In recent years, AI-generated images and videos have also seen tremendous improvements, to the point where the best AI visuals can be nearly indistinguishable from real ones.
Almost, but not entirely.
How to Identify AI-Generated Videos
Although some people shy away from using AI for text-to-video creations, AI-generated videos are popping up everywhere—from YouTube to TikTok, and across every social media platform you can think of. This means it’s essential for us to learn how to tell whether a video is the product of artificial intelligence.
Why does this matter? In today’s world, videos serve as a crucial source of information, and discerning what’s real versus what’s not is vital to avoid falling prey to misinformation.
There are still some clear indicators that can help us determine if a video is AI-generated. While these markers might not always hold true as AI continues to develop, they’re useful for now. Here are a few signs to watch out for:
1. Impossible Physics
A significant giveaway that a video may have been created by AI is the presence of actions that defy the laws of physics. While special effects may bend the rules in legitimate videos, it’s usually easy to differentiate between deliberate effects and unintentional mistakes that AI generates.
Examples include objects changing direction without any apparent force, unrealistic high jumps reminiscent of “The Matrix,” and liquids behaving like solids or vice versa. These inconsistencies occur in AI videos because the technology relies on data rather than a true understanding of physical laws.
2. Awkward Scene Transitions
If you’ve seen poorly made AI-generated videos, you know about jarring scene transitions. While some examples showcase comically bad cuts, even subtle transitions can offer clues. Watch for any figures or objects changing into something else in an unrealistic way.
An obvious tell is nonsensical cuts that confuse the viewer. Humans filming typically ensure that transitions maintain some narrative coherence, while AI lacks that concern, generating content frame by frame without applying a logical flow—unless prompted in detail.
3. Unnatural Human Movements and Expressions
Human expressions are incredibly nuanced, making them one of the best indicators when assessing whether a video has AI origins. We’re naturally observant of our peers and can easily notice subtle facial cues. However, generative AI struggles to mimic the complexity of human emotions and expressions.
In AI-generated content, these deficiencies manifest in both overt and subtle ways. Obvious signs include jerky arm or leg motions, while less apparent issues involve awkward mouth movements, unnatural blinking, and expressions that shift too abruptly.
4. Background Anomalies
Real-life videos capture entire scenes, while AI often neglects background details in favor of perfecting the main subject. This creates an opportunity to spot AI content by examining the areas beyond the focal point. Look for oddities—blurry textures, strange artifacts, or nonsensical elements may indicate AI involvement.
While AI attempts to create fitting backgrounds, it often fails to eliminate visual noise, which might result in flickering patterns or wobbling objects against a chaotic background.
5. Mismatched Actions and Emotions
If the video features a human, pay attention to discrepancies between actions and displayed emotions. Sometimes these inconsistencies can be subtle, but at times they’re glaring. For instance, if someone appears relaxed and happy while in a perilous situation, that’s a clear mismatch.
AI may handle the depiction of physical beings competently, but it struggles with facial expressions and mouth movements, often missing the delicate details required to convey genuine human emotions.
6. Illogical Narratives
Lastly, a hallmark of low-quality AI videos is the presence of nonsensical storylines. If you’re left puzzled about how or why certain events are occurring, it’s a likely sign that you’re watching AI-generated content.
A typical example might involve a celebrity engaging in absurd or improbable scenarios that simply don’t make sense, like Will Smith engaging in bizarre eating habits or Gordon Ramsay cooking in nonsensical ways that defy logic.
Overall: Trust Your Gut
While these indicators are helpful, the most crucial advice for identifying AI-generated videos is to trust your instincts. Humans have an innate ability to perceive subtleties—if something feels off, there’s a good chance it might be.
As attention spans dwindle, many may overlook signs of AI manipulation in videos, which poses a significant issue. Be vigilant about what you’re viewing, and remember: if it seems too polished to be real, it probably is.