WE WERE ALL AMAZED when the first image generating software was released, which was not all that long ago. DALL-E, Midjourney and others allowed you to create ‘art’ — even if the eventual generated pictures rode on the backs of millions of pieces of earlier, hand-drawn art. We’re still arguing to this day over how (or in some cases even whether) to compensate artists for this use of their work. It’s complicated when it’s difficult to positively identify component bits and pieces.
This is an argument that has almost been rendered (pun intended, I guess) irrelevant by the latest developments in picture-generating programs. The developers of image-generating AI have moved beyond still pictures into the realm of video. In ten or 15 minutes even children can now create very short videos (currently ten-15 seconds) that previously would have taken entire film and special-effects crews weeks to shoot. More advanced, professional versions allowed for the generation of longer videos that tell stories or act as complete music videos.
Sora presently being perhaps the best known, I thought I would download the base version and see if I could create some 15-second scenes that could pass as illustrations to some of my stories. The result is — well, it’s magic. At the risk of being redundant, Arthur C. Clarke once said, “Any sufficiently advanced technology is indistinguishable from magic.” Right again, Arthur, just as you were about communications satellites back in 1946.
I gave Sora a few commands (or “prompts,” as they are called), did what minimal revising I could with the consequences, and after a couple of hours ended up with short videos illustrating three different stories. All the videos had some problems, even after revising, because the editing tools are primitive in the basic version. But they were good enough to suggest that within a year or two anyone will be able to produce their own full-length movie at home, in the comfort of their own den-study-playroom. All that will be required is practice and sufficient processing power to render out the final results.
Not only does the software generate video, it can take different camera angles (another prompt), design characters, provide custom backgrounds or supply its own based on your little story, and add relevant, editable music and suggested dialogue appropriate to the scene. There are restrictions, of course. Nothing X-rated, no use of real people, etc. But if the history of the still-scene generators is any guide, users will quickly find ways around whatever guardrails the producers put in place. It has always been thus with software, and we have to be ready for it.
That most recent video of a “drug boat” being blown up by the US Coast Guard off the coast of Puerto Rico? Maybe it was done by a 17-year-old working out of the basement of his home in Langley, Virginia. The new song by Taylor Swift released as a New-Year surprise? Entirely generated by a combination of Sora, or Veo 3, or one of their competitors, combined with song generation by AI Sonos. That old saying “a picture doesn’t lie” is being rapidly followed by “film/video doesn’t lie” on the way to the ashcan of an increasingly outdated reality.
It's not just video generation from scratch, either. Have a look at Youtube videos from folks like The Artful Eye, where old artworks are brought to astonishing life. Who knew what lay unseen behind portraits of the beautiful and famous? AI fills in the space when the subjects move. Painted water now flows, candles burn, figures frozen in painted time now dance and smile and play instruments. Your deceased grandparents can, too. Just provide an old, grainy, black-and-white snapshot and AI will bring them to life once more. A lot to process there, and I don’t mean pixels.
I’m not sure where it will all lead. Nobody is. When software can put words in Lincoln’s mouth, there’s nothing to prevent it from putting words in the mouths of living politicians. Or actors. Or you, if there’s a picture of you anywhere on the Web. Legally, this software (which remember, is still in its infancy) opens up a new world of jurisprudence. I can see the day when AI legal software is asked to render a verdict on the legality of AI-generated video. No humans involved.
No humans necessary.
Prescott resident Alan Dean Foster is the author of 130 books. Follow him at AlanDeanFoster. com.
