The subject of AI rears its head this week with the release of SORA. This may not have got much press, but the implications are, quite frankly, earth-shaking. Most scientists working within and outside the AI world have been predicting that we'll have a full 'generalised' artificial intelligence or AGI in the next 60 to 100 years. This means one that understands the world and can respond to it with specialised code. With the advent of ChatGPT, those predictions have been coming down from 60 to 40 to perhaps 20 years. However, with the introduction of SORA, which is OpenAI's latest model for text-to-video conversion, industry insiders believe that OpenAI might well have that AGI already, and they're just working with it and trying to figure out the best way to present it to the world (very sensible if so).

So, how is text-to-video such a big leap? Well, in order to manage to create video, the AI needs to understand the world at a fairly in-depth level. How things fall, how things move, what happens when one thing goes behind another, etc. At the level of which SORA is working, the AI knows a hell of a lot about how the world works - too much to be considered just a text-to-video app. Some people will say that itís just a computer generating a bunch of frames one after another, which is like saying that human writing is just presenting one letter after another. Around this announcement are some fantastic presentations on what this release means. One of the other interesting aspects of this is that with an understanding of the world, generating 2D images becomes a hell of a lot more realistic. I've included a still from a video below (link to the video). We live in interesting times...

Click here to download issue 298 (high quality, 95Mb)

Click here to download issue 298 (smaller download, 45Mb)