New Video: Realtime Creativity / Explore Your Art with GPT-3 and DALL-E [GPT-X, DALL-E, and our Multimodal Future]
I’m really excited about today’s video. It’s one of those videos I really cherish and almost don’t want to let go of!
This video introduces you to the idea of real time creativity. This new phenomenon of generating creative work, will allow you to explore your art in greater detail than ever before.
YouTube Transcript (SPOILER WARNING)
Instant gratification is this really interesting social idea. But here’s the thing, we’re not used to thinking of instant gratification when it comes to creative work. That’s because creativity is normally a painstaking process and just simply takes a long time.
It’s not like you sit in front of Adobe illustrator and just “think” about an apple being made for you and it gets made.
You need to set up your layers, pick up different tools from the left, select and deselect various things, enter the colour you want, and more. Creating this simple, desired apple takes labour and time. This is a simple example, but every masterpiece is the sum of so many other pieces, tens of thousands of steps over time, maybe even more.
But in the world of multimodal and language AI models … life is just different. Even through GPT-3, I’ve become used to AI instantly writing on my behalf, without all those painful steps in between. For example, I previously made a video on writing press releases, this is normally a painstaking artisan writing process done by PR professionals. Watch me just give GPT-3 a headline and watch it instantly write a press release for me in response. There’s just no way a human could write something this coherent, this fast in a single draft in real time.
This is just a new way of formulating creative work altogether. In the past, you just wouldn’t instantly generate an architectural design, or instantly generate a heartfelt song, or an entire video sequence altogether, but here we are. I call this new phenomenon realtime creativity.
But if we can agree, yes, creativity may become a lot less laborious and maybe even instantly gratifying, through multimodal AI models like DALL-E, how will this change how we create?
Well, for one thing, I just think since you’ll be less attached to the labour of generating stuff, you’ll be more bold with what you do with it. You’ll be willing to edit, mold, or shape it in radical ways to your true vision, or look for something new altogether. In other words, I just think you’ll experiment a lot more. The nature of your work will no longer be in creating a design but rather all the experimentation you did after the design was already made for you by the computer generating it for you symbiotically.
Going back to the multimodal photo editor concept, let’s say I already have my scene and used a multimodal AI model to generate this apple, this entire process took just a few seconds. Now, my purpose of this scene is to make an apple which looks appetizing, that’s the goal here.
But, to be honest, I’m not satisfied with what I’m looking at so far. So, let’s try … let’s say I actually want the apple just to be larger, so it draws people’s attention to it. That’s better, but actually, I want the apple to be larger and a different variation than the current one. I just don’t like this one, for whatever reason. There we go, this is good … this is a decent one.
You can see I'm going deeper into exploring this illustration. Since I wasn’t the one who made it from scratch, I didn’t just settle on the first version.
But now that I think about it … what if this scene was more of a close up to the apple? Or maybe it should already be bitten a little bit with a different orientation and leaf to increase the viewer’s appetite for it. Or actually, what if the whole scene was that same apple becoming an apple pie instead? Would that be more appetizing to viewers and make them hungry? Let’s see. Hmm. The cool part is, in a few seconds I was able to test all of this!
So, what’s the point here? I often think my work could be a lot better if I wasn’t so vested and spent so much time creating everything from scratch. I wish I could sit on top of work that’s already been done for me, but be able to manipulate it down in significantly foundational ways, to really explore the essence of what I’m saying and how it’s being said. But also, I want to be liberated and have more opportunity to ask what if? What if this was a different colour, what if it took place somewhere else?
The Key Idea:
Explore. Your. Art. Realtime creativity will enable us to leverage multimodal AI models to experiment with our ideas deeper. It’s also about exploring their potential as alternative realities. Learn to go deeper, but also see how your ideas would look from entirely new angles. Ask meaningful questions about the nature of your creations. Ask what if?