11 months ago, we all shared a laugh at stupid, old generative AI producing videos.
Whenever I hear someone say how bad generative AI is at doing something, they are not paying attention. It is like making fun of your kid for not knowing how to do something. They don’t know how to do something… yet. AI is the same way.
“Gen AI is at its worst… today.”
Hans Scharler
Today, OpenAI announced Sora. Sora is an AI model that can create realistic and imaginative scenes from text prompts. The OpenAI blog post is filled with examples of AI-generated video clips. They promise that they are unedited.
So, in 11 months, we went from a horrifying clip of Will Smith (not the Oscars one, the AI-generated one) to basically one-minute of passable video. Video that existed moments before, encapsulated in a few English words.
Reviving the Wooly Mammoth
Let’s say you wanted some stock footage of wooly mammoths running through a field. All you have to do is send a prompt to the Sora model. A prompt is just a set of instructions, descriptions, and intentions that will elicit an output from an AI model like Sora.
Here’s the Sora AI prompt:
Several giant wooly mammoths approach treading through a snowy meadow, their long wooly fur lightly blows in the wind as they walk, snow covered trees and dramatic snow capped mountains in the distance, mid afternoon light with wispy clouds and a sun high in the distance creates a warm glow, the low camera view is stunning capturing the large furry mammal with beautiful photography, depth of field.So, what else do you think AI is terrible at? What else are we making fun of? When are we going to address this head on?
Majestic mastodons sure. But how good has it gotten at generating Will Smith eating spaghetti!?
Apologizes. I missed your comment. Yes, this is the new Turing Test for generative models. I will let you know when I can try it.