New Video: Five Ways to Make New Things with Multimodal AI [GPT-X, DALL-E, and our Multimodal Future]
Welcome back! How was your weekend?
Today’s video answers a very important question that may have been bubbling in the back of your head:
If everyone is using the same AI tools, how will you make something different that stands out?
I loved making this video and really hope the predictions I make in it stand the test of time. Let me know your thought on the video in the YouTube comments or comment section below.
YouTube Transcript (SPOILER WARNING)
You might be wondering, if everyone has access to the same tools which are pretty easy to use, will everybody be a creator? Won’t it get really saturated and how will everyone compete?
Also, won’t the same models generate the same things for everyone?
Will these models ever come up with new things?
Where will originality come from?
My hunch is we will be able to actually train and configure AI models to generate novel things, but in this video, I am going to quickly share in the future to keep things fresh with your multimodal AI creative work.
1) Mixing and texturing
To combat a lack of originality, you’ve already seen my earlier video in this series on mixing and texturing. So, my recommendation is if you want to make stuff that the world has never seen before, mix completely new and unique things using multimodal AI.
2) Experimentation and exploration
In a later video, I’ll also be talking about the importance of experimentation. With enough rapid experimentation and deep exploration after your concept has already been generated - you can discover something entirely new and different.
3) Systems design
I think systems design is a massive opportunity for multimodal AI creatives.
Imagine multimodal AI not just generating a graphic for you, but working with it to generate your entire creative campaign. In the fashion case, everything from the music, to designing the clothes, the event programme brochure you would hand out, and even the marketing campaigns to promote it, could all theoretically be generated by multimodal AI.
By designing your campaign at a systems level, which means holistically but also looking how at how each piece is integrated, you will create something entirely new and refreshing with greater consistency across the board.
4) Context and framing
There’s also something to be said about how you communicate your project’s unique emotions and identity from the very beginning. Which is where mood boards come in. Mood boards are a really important planning and communication tool for creativity.
In fashion design, creatives often start with some kind of mood board first - this mood board may contain colour palettes, textures, quotes, and other fashion items that the designer finds inspiring and relevant to the specific line they’re working on. Imagine giving a one-of-a-kind of mood board to a multimodal AI system and getting back new clothing designs the world hasn’t seen before. This has been my number one request for GPT-3 and I hope we can incorporate mood board configuration settings for creative multimodal AI tools in the future.
Concept art and pre-visualization is another way to explore a project before it is formally made, in the future, I think all of these planning tools will be powered by multimodal AI models.
By creating more context and framing our ideas better through unique, highly specific mood boards and concept art, I think we can generate greater, more novel outputs from multimodal AI models.
But, what makes a good mood board? This brings us to our final point.
5) Great Taste
So, for me personally, I’ve always believed creativity is first and foremost about taste. Two people can have access to the same music production tools, but I think there’s something distinctive about music and beats made from some people over others. The best stuff rises to the top and really it’s the taste makers of the creative class who drive everything forward.
Unfortunately, there’s not much I can say about taste beyond that … from what I’ve heard, the one thing about it, is that some people have it and some people don’t. I’m not sure if this is the kind of thing which can be taught in schools.
So, in short, where will novelty come from? In my view, through multimodal mixing and texturing, experimentation or exploration, higher levels of system design, greater creative context and concept discovery through things like moodboards, and finally through very talented individuals with the highest levels of creative taste.