Sunday, December 22, 2024
Business

OpenAI's new 'Sora' text-to-video tool unveiled by Sam Altman

Any fears that OpenAI had lost its magic after November’s crisis over the firing (and subsequent rehiring) of Sam Altman look like they can now be emphatically put to bed. 

On Thursday, the company’s CEO revealed a new AI tool that, using nothing more than a brief text prompt, creates videos up to 60 seconds long so realistic the average person would struggle to guess they were generated by a computer.

Whether it’s stylized imagery of a California town during the 1840s gold rush or photorealistic renderings of golden retriever puppies playing in the snow, the lifelike animation surpassed anything from other text-to-video competitors like Runway and ModelScope.

“This is one of the most convincing AI-generated videos I’ve ever seen,” said Marques Brownlee, a consumer tech YouTuber with over 18 million followers. “This does feel like another ChatGPT / DALL-E moment for AI.” 

A few years ago the height of AI videos was a deepfake Tom Cruise, but those took time to carefully splice the actor’s countenance onto that of an impersonator.

Fully-generated videos by comparison suffered from glaring limitations, such as one that went viral only 11 months ago that showed Will Smith shoveling down platefuls of pasta.

“It’s truly the stuff of nightmares—which perfectly demonstrates the current, struggling reality of AI video generators,” Futurism’s Victor Tangermann wrote at the time.

What impresses most about OpenAI’s Sora is its ability to simulate the complicated physics of motion while simultaneously showing a baffling capacity to mimic real-world lighting effects.

Typically only high-performance game engines like Epic’s Unreal Engine 5 are that advanced, leading some users to wonder whether the latter was somehow employed. 

To prove the handful of examples on OpenAI’s website were not cherry-picked out of millions of less impressive results, Altman took to social media on Thursday to gather random and often outlandish suggestions for themes his Sora would then transform into videos.

“Don’t hold back on the detail or difficulty!” he urged

OpenAI’s stunning comeback

Sora marks a stunning comeback for a company nearly torn apart by its biggest-ever internal crisis just three months ago.

The termination of Altman followed by his almost immediate rehiring exposed a deep governance rift between OpenAI’s non-profit origin and the for-profit model backed by large shareholder Microsoft.

Speculation ran rampant over whether the chaos at OpenAI would spark a brain drain as staff chose sides over whether to commercialize the technology as fast as possible or be more mindful of the original AI safety-driven mission. (OpenAI co-founder Elon Musk took the opportunity Sora provided to remind the CEO of his betrayal.)

There is no shortage of companies eager to profit from a potential exodus of talent, as Salesforce founder Marc Benioff proved.

It didn’t help confidence either that Andrej Karpathy, a high-profile member of the founding OpenAI team, revealed earlier this week that he had once again left the company after rejoining only a year prior. 

This suggested the team may have lost urgency and focus in the wake of November’s debacle.

Innovation speed

The arrival of a tool as powerful as Sora only months after previous sophomoric attempts like the Will Smith video helps explain both the pace of innovation in the field as well as the ongoing boom in AI stocks like Nvidia.

The premier vendor of AI training chips rallied 50% in the past six weeks alone, in the process eclipsing both Amazon and Alphabet as the world’s fourth largest company by market cap. 

Even as the share price of companies like Nvidia spike, the technology has triggered deep-seated fears that humans may find themselves the lesser species should people like Altman succeed in creating artificial general Intelligence.

When confronted with the question last month at the World Economic Forum in Davos, the OpenAI boss didn’t have a convincing answer, either. 

While it may have been partly tongue-in-cheek, the Wall Street Journal’s executive editor for video concisely summed up what Sora might mean for the value of her labor.

“It was nice knowing you all,” Joanna Stern posted on Thursday. “Please tell your grandchildren about my videos and the lengths we went to actually record them.” 

Subscribe to the Eye on AI newsletter to stay abreast of how AI is shaping the future of business. Sign up for free.


source

Leave a Reply

Your email address will not be published. Required fields are marked *