OpenAI sora text to video generation model

OpenAI’s Sora Sparks Memes, Fear, And Horror

2 min read
Disclaimer

This article is for general information purposes only and isn’t intended to be financial product advice. You should always obtain your own independent advice before making any financial decisions. The Chainsaw and its contributors aren’t liable for any decisions based on this content.

Share

Follow

Last week, OpenAI introduced Sora, the company’s latest text-to-video generation model that sparked fascination and horror on the internet.

Enter a text prompt and describe your desired scene, and Sora is able to create videos of up to 60 seconds based on said prompt. On one hand, techno-positive individuals marvelled at Sora’s ability to generate highly-photorealistic clips with complex camera motion. On the other, creatives expressed anxiety over the very real possibility of Sora threatening the filmmaking business.

Sora and misinformation

Sora’s debut over the weekend, unsurprisingly, reignited discussion and fear over the proliferation of AI-generated deepfakes and misinformation.

“With the release of OpenAI Sora, I IMPLORE YOU, I am literally begging you, to be sceptical of what you see on the internet,” wrote internet personality Tank Sinatra on Threads.

“I understand I’ve spent 20 years staring at a screen and maybe I can tell things are fake where others can’t, but if you see a video of someone saying something so outlandish it could start a war or ruin someone’s life, please, at least try to find another source for it before you go neck deep into the conspiracy vortex.” he added.

… and Sinatra is not wrong. If AI-generated deepfake images and videos aren’t concerning enough, we are about to enter a new age of rampant misinformation.  

Mistakes still exist

As impressive as Sora appears to be, it still made mistakes. One user cited a ‘failed’ video as an example, which showed a spilled drink. However, the video contains multiple anomalies: Sora first depicted liquid in the glass to leak. Only after the drink finished leaking, the glass tumbled over. 

“It knows what a spilled drink looks like – but isn’t sure yet how drinks actually spill.” 

In other words, Sora still doesn’t possess ‘common sense’ understanding that most of us humans have.  

Source

“… None of our video and image models actually understand images. They know what are likely to be correct images, but there’s no understanding of the relationship between entities, physics, etc,” explained a product manager at Meta. 

But for now… memes

Sora’s AI-generated videos went so viral that some skeptics have also taken the opportunity to poke fun at the tool.

“Prompt: Australian man wrongfully arrested while eating a succulent Chinese meal,” wrote a user on X. Can you guess what Sora created?!

Here’s a second Sora masterpiece: “A young man, ginger hair, sings a song in front of various urban backgrounds, 80s hairstyle and outfit, wild dancing gyrations, background dancers, 80s video resolution, photorealistic, pop video.”

You’ve been Rick-rolled.