Five Simple Ways To Determine
Whether A Video Is Generative AI
Most people use the internet to view sports highlight videos, breaking news clips, death-defying stunts, cat memes, and videos of the day. Videos actually account for almost 80% of all internet traffic because users prefer them over other forms of content.
In the upcoming years, computer-produced video is anticipated to drive internet video consumption even higher due to the introduction of generative AI. According to Grand View Research, the market for AI-generated video was estimated to be worth $555 million in 2023 and is expected to nearly quadruple to $1.96 billion in the following five years.
Videos that have been combined by AI have been making news and causing problems even prior to that kind of rapid expansion. In particular, one of the most well-known issues with generative AI technology is deepfake videos. While some deepfakes have been nothing more than lighthearted parodies, other more ominous videos made with artificial intelligence (AI) from text and still images have included fake footage.
Authorities have found that several of those AI-generated videos are in violation of numerous privacy laws, legislation pertaining to sexual or child abuse, and even national security directives concerning election interference. The importance of this issue has led California to pass the first anti-deepfake laws in the country, however those laws will probably face challenges to their constitutionality.
For the time being, regulations prohibiting AI-generated videos appear hazy at best, so we must educate ourselves on what to look for.
Here are five ways to help determine whether the next video clip you view was produced using artificial intelligence (AI), which is a continuation of my articles from last week on how to identify AI-generated written content and photos.
Here are five ways to help determine whether the next video clip you view was produced using artificial intelligence (AI), which is a continuation of my articles from last week on how to identify AI-generated written content and photos.
There may be misalignments between lip movements and spoken phrases in these AI-generated videos, particularly in deepfakes. It’s similar to watching foreign movies with an English overdub.
With one exception: in this case, the audio performer and the on-screen actor are using the same language.
There may be a clue that this discrepancy is an AI replica.
Blinking and Expressions on the Face
In generative AI videos, it’s common for humans to blink artificially. In the real world, humans blink at erratic intervals that are connected to spoken cues during discussions.
Furthermore, if you look for it, their limited or unduly smooth facial motions can be distinguished.
Though AI programming hasn’t yet grasped these subtleties, it will in due course, which is why AI-video detection technologies are essential.
The first tool to take into account is the Deepware scanner, a user-friendly detector that focuses on identifying deepfake movies by examining minute variations in facial expressions and frame-by-frame anomalies.
WeVerify provides its handy InVid Chrome plugin, which allows you to have an AI detector toolbox active on your browser at all times. This is another helpful resource. The nonprofit Poynter Institute for Journalism Ethics has recommended the plugin, which took top place in the 2021 U.S.
Paris Tech competition.
It’s a best practice to employ many tools, which is why we are releasing multiple AI detection tools. Make it a practice to confirm the first tool’s findings with a second one.
The current generation of generative AI platforms struggles to reliably depict background features and depth of scale in movies, much as they do with AI-generated photos.
AI movies frequently contain small aberrations that can show up as flickering, pattern noise, or strange changes in the surrounding landscape. These constructions are usually the result of existing rendering process restrictions in AI. Anticipate those to get better with time.
Videos produced by AI are remarkably adept at gradually incorporating weird facial expressions, clunky speech, and strange microgestures. The result of all these little errors is an inconsistency between the spoken emotions and the outwardly manifested responses.
Once more, the discrepancy is minor but frequently sufficient to alert you to the possibility that there is something off with the video.
The ability to separate true videos from fake ones is critical when you consider the power of video messaging. The brain processes video images 60,000 times faster than text and, as a result, 95% of a video message tends to be retained compared to just 10% of written text.
If the proliferation of online video and generative AI video in particular continues as projected, one day we may be forced into a state where “seeing is believing,” even if what we see is completely AI fabricated.
Discover more from Postbox Live
Subscribe to get the latest posts sent to your email.
