The Middle Table AI Challenge: The Why and the How

I asked ChatGPT to write the intro to this blog. Here’s what it gave me.

AI is revolutionising video production, offering powerful tools for automation and stunning effects. Yet, it sparks controversy: Is it enhancing creativity or risking the loss of human artistry? As technology advances, the balance between innovation and tradition hangs in the balance.

I mean, there’s nothing wrong with that, is there? It sets out the scene nicely, summarising the key arguments in a succinct, punchy way. And it gave me it in under a second, with a single request. AI is impressive, for sure. But, having scratched away at its super-slick surface for the best part of a year, what I really wanted to know was: can AI make anything great? And how easy is it to use?

Those were the questions that inspired the AI Challenge, a task we recently embarked on here at Middle Table. We’re just the type of creative agency that finds itself confronting the tantalising and terrifying prospects that AI offers, so why not grasp the nettle?

We settled on a plan. The team would get an ordinary brief - one week to make an ad for a futuristic smartwatch, the Xaltris A-200 - and throughout every aspect of production they had to rely on the brains of any AI software they could find. Generate ideas and form scripts using ChatGPT or its equivalents; storyboard with Storyboard Hero or Storyboarder; produce images and video using Midjourney, Runway, Adobe Firefly, Kaiber. Voiceover, music and editing would have to be covered by the likes of ElevenLabs, Suno and Veed. Xaltris’s founder, Arty Fishial, should be included in the video but the team had to use AI to adapt an existing, unrelated clip into a bespoke soundbite. Essentially, anything that AI can do, it had to do. Ordinary programs could be used only in exceptional circumstances.

In the process of production, the team would get a sort of crash course in learning the ropes of AI, testing both its and their own strengths and weaknesses. Which programs are best? How important are the intricacies of prompting? And is it even possible for an AI character to look normal while eating, drinking, walking, gesturing…?

There are, of course, enough legal and ethical concerns to floor this film before it even leaves the paper (oh you’re rendering voiceover artists redundant, I see) and we’d chew over those as we went along. But, for now, the only barrier to making a masterpiece was the creative capability of this uneasy human-machine partnership.

Stay tuned for how it went…