A little over a year ago, specifically 381 days ago, I made the prediction that in the future, we would see the following:
Artificial intelligence gets good enough to generate content — TV shows, music, books.
But good enough for what?
Good enough for each of us. Each of us gets a custom stream of entertainment, based on our previous preferences… based on how our eyes dilate… based on whether we keep watching.
Each of us is served with the perfect content, just for us, just for that moment. Familiar enough… with the right amount of surprise to keep us fascinated and perfectly pleased.
I wrote that after seeing a series of AI apps that could generate realistic human faces… cat pictures… even functional music videos.
Well, a year has passed. And damn.
Maybe you saw the announcement three days ago about Dall-E-2.
Dall-E-1 appeared a year ago. It’s an AI thing that generates images from text descriptions. “A baby daikon radish in a tutu walking a dog” turns into a picture of just that.
Thing is, Dall-E-1 produced stiff, flat, lifeless pictures. I saw it last year. It was interesting but not very impressive.
But Dall-E-2? The one that they just got released three days ago?
It’s creative. Innovative. Even beautiful. The images it produces are understandable and yet surprising. Good enough to keep us, or at least me, fascinated and perfectly pleased.
And it all happened so fast. In one year, the state of the art in AI went from stiff wooden clipart to something bordering on real art or at least high-level graphic design.
My point being:
After the initial rush and push and fear that AI is coming to take your job — if you’re a copywriter, at least — it seems the noise died down. So it is always. We tire of the news, even when it becomes more relevant.
Instead, a complacent attitude set in. “Yeah yeah, AI will never matter for real copywriters. Or at least not any time soon. Human beings want human interaction, personality, warmth, insight.”
The initial AI-will-take-your-copywriting-job discussion came after GPT-3 was released a couple years ago.
But I’m much more impressed by this Dall-E-2. Take a look at the link below and see if you agree. It’s a bunch of Dall-E-2 generated images from cryptic, flowery Twitter bios (such as “surfin’ and slingin'”).
This Dall-E-2 technology does not output text. But when something similar comes for text generation from short prompts…
Well, all I’m saying is, start thinking of how to take your writing skills or your copy-based business… and turn it into something that will still be relevant in a year or maybe just six months from now, when the new update to GPT-3 comes out.
Or don’t.
You can also just wait those 6 months, and you will probably be able to ask GPT-4 to answer that question for you — better than you can, leaving you both impressed and a little humiliated at the superior intelligence and insight this thing has.
Anyways, that’s my new prediction.
But maybe I’m just blowing it all out of proportion. If you want to see Dall-E-2 in action for yourself, and see if I’m exaggerating, take a look below.
But before you go, sign up to my email newsletter. I still have six months before the Apocalypse to write you some human-generated content.
Anyways, here’s Dall-E-2:
https://twitter.com/nickcammarata/status/1511861061988892675