AI is getting better and better every day at generating an image from a few words, with public AI image generators such as DALL-E 2 and Stable Diffusion.
Now, Meta researchers are taking AI a step further: they’re using it to make videos out of text.
Meta CEO Mark Zuckerberg on Thursday tweeted about the study, called Make-A-Video, with a 20-second clip that compiled the texts the Meta researchers used into very short videos.
The lyrics included words like: “A teddy bear painting a self-portrait,” “A spaceship landing on Mars,” “A little sloth in a knit hat trying to work on a laptop,” and “A robot gliding over a wave in the ocean”.
The videos for each text are only a few seconds long, and generally show what the text suggests, in a low resolution and somewhat rushed style.
However, it demonstrates a new direction artificial intelligence research is taking as systems get better and better at generating images from words.
However, if the technology is eventually released widely, it could raise many of the same concerns raised by text-to-image systems, such as that the technology could be used to spread misinformation through videos.
In his Facebook post, Zuckerberg pointed out how complicated it is to generate an animated image from a few words.