James Ross, our Director of Strategy shares his thoughts on the latest AI model from Open AI.
So what is Sora?
It’s the newest Sorra AI model from the teams at OpenAI that creates video from text. It is a combination of realistic and imaginative scenes that come from text-based instructions.
A newest model sounds modest in comparison to what we received. This wasn’t an iPhone 11 to iPhone 12 model update or refresh. This was a mind-blowing, giant step of improvement in creative output from AI.
What is truly impressive is the use of physics, multitude of filming angles and unlimited concepts that it can pull together.
Check out this video with a prompt based on a stylish woman walking through Tokyo:
Sora led me to have 3 thoughts about its (and AI’s) future:
1. The AI learning curve
How fast is AI improving? It is just amazing. It is the fastest improving technology the world has ever seen.
If you still think AI isn’t going to be a very active part of your role, every single day, in the next couple of years you need to wake up and get clued up.
Just watching these types of videos should be more than enough evidence to show what is coming.
It is going to completely reshape teams, agencies, marketing departments. Not in a ‘robots will take our jobs’ way but in how we can use our time to be more effective, creative and strategic.
2. Mega accessibility leading to unlimited use cases
For now, Sora are inviting some filmmakers, visual artists and designers to further test it.
When the time comes, that access will be opened up and more of us may use it and the general use cases for it will be far and wide.
For a long time we’ve had our camera phone but if you’re like me turning my home footage into studio production quality isn’t my strong point. Nor do I tend to bring with me a film crew, lighting equipment, CGI specialists, drones…you get the point.
I’m sure even as you watched the video you started thinking about how you might use it. Or how the company you work for might use it? Perhaps a marketing application like dynamic creative, or for start-ups to get a leg up with stock footage.
It really will be fun to watch how tools like this get used and applied in the everyday. This was my favourite example from Sora, historical footage of the Californian gold rush…
3. Licensing, safety, physics challenges
How will Sora licence the footage, will it be universally available to use? Will it be as accessible as I dream it to be? I’m not sure.
The question of licensing and usage is interesting and yet still a little murky. I foresee some companies (especially challenger brands) being more inclined to use Sora and AI more actively whilst larger, stable brands coming under fire for it with the knowledge that they could be paying for models for example.
Safety is mentioned on the Sora website, front and centre. The question of monitoring AI and the appropriate usage of it should be the highest priority. There will continue to be question marks over the quality of detection systems and classifiers; particularly in the early stages of release.
Finally, the quality of the tool will still be a challenge. AI is still learning human physics. Even watching the videos if you stare at the legs of the model you’ll see jolting, overlapping legs more regularly than you might expect. With the learning curve AI has been on I can’t see this being an issue for long.
So the AI conversation continues. It’s moving fast and I’m watching in total fascination of what is to come. I look forward to getting hands on with Sora and as we move forward developing use cases for the marketing world.
