The latest artificial intelligence news is about OpenAI’s Sora, a program that creates short videos based on text prompts. OpenAI isn’t making Sora available to the general public, but it has released samples to give people an idea of where the organization is going. The online New York Times (NYT) has shown some of the samples.
One of them is a short clip of wooly mammoths galloping in a snowy meadow, with evergreen trees behind them, towered over by snow-capped mountains. The mammoths are kicking up clouds of powdered snow. Their fur is flapping as they run.
To me, the video seemed like what a movie might look like with painstaking CGI, except it was generated by a text description no more detailed than the one I wrote in the last paragraph.
In another sample, a Japanese couple is followed from behind and above as they stroll along a pedestrian-only street in Tokyo or, more precisely, an imagining of Tokyo. They pass shops on the right, the road they stroll upon is lined on the left by blooming cherry blossom trees and a bridge arches over a river in the background, with a cityscape beyond it.
One the NYT article didn’t show was a video of a turtle riding a bicycle underwater on the floor of an ocean or lake.
There’s also a weird, surreal-looking video of fish swimming around (in a coral reef?), all looking like they’re made of brightly colored construction paper.
OpenAI has also brought us ChatGPT and the DALL-E text-to-image generator. OpenAI says it is releasing the samples for feedback from the public concerning possible uses and abuses of the technology and ideas for making it safe.
There’s the usual concerns that the technology can put creators out of work and that the images could be used to spread misinformation.
Or it could further erode trust. In 1969, I watched Neal Armstrong and Buzz Aldrin climb out of the LEM onto the moon, filmed by a remote-controlled camera on a mechanical arm swung down from the LEM. I knew it wasn’t faked. But, the same year, “2001: A Space Odyssey” started convincing people it could have been faked, because didn’t Kubrick do just about the same in Hollywood studios? Next time I see a video of wooly mammoths frolicking around a snowy meadow like giant puppies, I’m going to think, “Oh, it’s a remake of the OpenAI classic.”
But my strongest reaction to it all is — I want this. I want dogs playing poker nonstop. I want it badly.
I want to tell a computer, “I’d like a herd of rampaging emus running wild through the streets of Paris. No, not just any street, up the Champs-Élysées to the Arc de Triomphe and back, harassing tourists and randomly scattered mimes all along the way. The mimes are offended and hurl insults at the birds in French, but the birds only mock them silently in return. Emus don’t speak French.”
Or, “I want to see a drunk polar bear wear a party hat and ride a unicycle on Arctic ice, slip and slide, but never fall off.” Please.
Six words: “Giant, mutant honey badger attacks Tokyo.”
I would soon get tired of videos featuring wildlife doing strange things. I might turn to re-imaginings of plays, movies and TV shows. How about “Hamlet” where all the characters are played by expensive rounds of cheeses? Will the computer ever refuse a request? “Cheeses don’t talk!” Or would it always find a workaround? “Alas, poor Yorick! I knew him, Horatio. He was a Gouda.”
Before I found the OpenAI Sora story, I’d planned to spend time here sharing the wit and wisdom of my mother. My mother was raised by an immigrant English school marm, and I think it warped her for life. Anyway, here’s a couple of samples:
“Never insult someone behind their back, when you can insult them to their face with more effect.”
“It’s better to be clever and manipulative than to know stuff. There are always people who know stuff who you can trick into doing what you want done.”
“Never send someone to milk the cows who doesn’t know the difference between a cow and a bull.” Note how this appears to contradict the previous notion somewhat. But, really it doesn’t, because you aren’t the one who’s gonna milk the bull. Use common sense!
Dr. Wes is the Real Change Circulation Specialist, but, in addition to his skills with a spreadsheet, he writes this weekly column about whatever recent going-ons caught his attention. Dr. Wes has contributed to the paper since 1994. Curious about his process or have a response to one of his columns? Connect with him at [email protected].
Read more of the Feb. 21–27, 2024 issue.