This is an I-Heart podcast.

Guaranteed Human.

Cool Zone Media.

Hello and welcome to Better Offline.

As usual, I'm your host, Ed Zittron.

A few months ago, OpenAI, OpenAI showed off SORA, a product that can generate videos based on a short text prompt,

kind of like ChatGPT does for text or Dali does for images.

These videos, which are usually no more than 60 seconds long, can at times

seem impressive until you notice a little detail that breaks the entire facade, like in a video

where a cat wakes up its owner, but the owner's arm appears to be part of the cushion, and the

cat's pore explodes out of its arm like an amoeba. Reactions to Sauras AI-generative videos and indeed the existence of the model itself have

ranged from kind of a breathless hype to genuine fear that this will be used to replace video producers,

in that it can create reality adjacent videos that for a few seconds kind of seem real, especially

in the case in some of OpenAI's hand-picked demo videos.

Yet even in these hand-picked Sora outputs, you'll find these weird little things that

immediately shatter the illusion, like one where a woman's legs awkwardly shuffle, then somehow

switch sides as she walks around, or blobs of people merging in the background of images.

These are, on some level, genuinely remarkable technological achievements, until you consider

that what they are and what they might do, and that there are problems in them that run

through the entire fabric of artificial intelligence.

A little over a month after SORA was announced, Open AI would debut a series of short

films, including one called Airhead, where filmmaker's shy kids told the story of a man with

...