The next wave of AI disruption is here.
On Thursday, OpenAI, the maker of ChatGPT, released a new product that can create incredibly lifelike videos based on nothing more than the words that a user types into a box. The video clips—which include an AI-generated woman walking down the neon-lined streets of Tokyo, astronauts wearing wool knitted helmets, and a young man reading a book on a cloud—raise the stakes in the already-highly-competitive AI industry and herald fast-approaching changes that are almost certain to roil the entertainment and content creation industries.
“If anyone thought the rate of progress in AI was going to slow down, we’re now seeing daily examples of exactly the opposite,” Box CEO Aaron Levie tweeted alongside a video of a clip generated by Sora.
Sora, as OpenAI is calling its new text-to-video product, isn’t exactly the first of its kind. Runway, an AI startup that has raised $263 million in funding, hosts an entire film festival dedicated to artists creating movies with its text-to-video technology. There’s also Midjourney, which offers high quality images from text.
But what’s most jarring about the videos OpenAI published Thursday is the fidelity and photo-realistic detail created by the company’s AI models, from the coiled strands of hair of the person reading a book in one clip to the scattered puddles on the ground in the Tokyo scene. The idea of creating such realistic video from text prompts was not only impossible a few years ago, but almost unimaginable with the tools and models at the time.
“This was one of those things you tell yourself is coming and you think you are ready for it and couldn’t possibly be surprised by it, but then you see it and don’t quite believe it and you’re not sure why you didn’t think you’d be surprised,” OpenAI COO Brad Lightcap tweeted.
The product underscores the exponential pace at which AI technology is advancing, something that futurists like Ray Kurzweil have long discussed. In “The AI Revolution: The Road to Superintelligence,” famed tech writer Tim Urban riffs off Kurzweil who predicted that the 21st century “will achieve 1,000 times the progress of the 20th century.”
OpenAI’s goal is nothing less than to create artificial general intelligence, a version of AI that can perform most tasks as well as or better than humans. Not everyone is convinced that such superintelligence is even possible. But the rapid pace at which AI is advancing is clearly visible in the flurry of new products. It was only 14 months ago that ChatGPT transformed the tech industry by launching a bot that can write poetry, term papers and conduct conversations nearly impossible to distinguish from a conversation with a human.
Just last year, Midjourney’s AI-generated video was still being criticized for putting additional fingers on a hand. OpenAI’s Sora has made that problem seem as antiquated as a modem’s dial tone.
For the AI startups that were already focused on video, OpenAI’s latest product is a shot across the bow.
Runway, a startup founded in 2018, has enjoyed a distinct advantage, having dedicated years to refining their models before other startups joined in. Until now, no rival AI text-to-video startup has posed a threat to Runway’s product. It’s worth mentioning Runway is already being used by Hollywood, a movie using their product won a whopping 7 oscars, while OpenAI’s Sora is only making short videos—for now.
“Game on,” tweeted Runway CEO Cristobal Valenzuela on Thursday, following OpenAI’s annoucement.
Perhaps the biggest obstacle for OpenAI in its video push is the need to defend its turf and continue to advance on so many different fronts, with competition heating up against ChatGPT and its image-generating tool DALL-E. OpenAI CEO Sam Altman has proven adept at staying ahead of the pack so far, raising $13 billion from Microsoft and reportedly seeking to raise even more money to develop hardware infrastructure capable of powering his vision.
As impressive as they are, the videos released by OpenAI on Thursday represent a limited, hand-picked exhibit of the technology’s best work. How well the technology performs overall will determine the inroads OpenAI is able to make into Hollywood as it seeks to broaden its addressable market and take on AI rivals. Game on.