There are a number of AI text-to-video tools out there, and while the technology is impressive, there’s always something about the end result that feels off. It takes a while to pinpoint the problem, but it all comes down to one thing: the uncanny valley effect.
While you can use some AI-powered tools to create visual effects in your video projects, it is not recommended to use AI on its own to create video footage because it looks too weird.
The Biggest Problem with AI Text-to-Video Tools
AI video generation has come a long way in a short period of time thanks to advances in Deep Learning. In 2023, when AI video generation exploded, you might remember this clip of Will Smith eating spaghetti going viral. While this type of technology was groundbreaking at the time, there’s no denying how unnatural and unsightly it looked.
In 2024, these AI video tools are getting better, producing smoother images and more realistic motion. Take a look at the difference between videos created with Runway Gen-2 in 2023 and videos OpenAI released in 2024 to showcase Sora AI. Sora isn’t available for public use yet, but the quality looks promising:
Despite the improvements, the results are still inconclusive. First, Sora isn't available yet, so we're still stuck with less sophisticated generators that can produce creepy results like Will Smith's spaghetti video.
Just watch the video created with PixVerse using the prompt "A person walking through a park on a sunny day, smiling and waving at the camera. Birds are flying overhead, and trees are swaying gently in the breeze" .
The first two seconds look fine, until the person's fingers, hair, and face start to vanish into thin air! Even as more advanced tools like Sora come along and give us more accurate and beautiful videos, there's still something unsettling about AI-generated people and landscapes.
While older models often produced videos that clearly showed off the AI, like claymation-style images, the improvements from newer tools look almost too perfect.
Unnatural, uncomfortable, and lifeless. This is exactly the uncanny valley effect - human, but not quite.
No matter how good these tools are, the uncanny valley effect still exists. Unless you’re going for an abstract, surreal style that looks like something you’ve only seen in your dreams, you shouldn’t rely on AI text-to-video tools for any of your video projects.