AI meets OMG

July 26, 2022

Having such great clients in the innovation space means TBG often has the opportunity to see amazing technological breakthroughs, as they are being created. It’s an honor to help share those stories, and witness the impact of these solutions, first-hand. Because from robotics to space research to cyber, the evolution of our clients’ worlds is changing all of our worlds, every day.

Perhaps one of the more ubiquitous fields of study for our clients – and, as marketers, our entire agency – is artificial intelligence. You’ve heard all about AI. It can be used to help recommend a movie for tonight, or help a vehicle drive itself on the most fuel-efficient route. It can be used to predict drought or a future purchase on Amazon.

In other words, it’s growing, it’s everywhere, and it continues to give technology awesome new capabilities across nearly every field. Seems like every day something is developed that makes you think, “machines can really do that?”

For starters, who’s a better artist – you, or your computer? Google Research is developing Imagen, an AI-powered text-to-image diffusion model that creates photorealistic images based on inputs from plain language. In other words, tell the AI to create an image of something – anything, the parameters are up to you – and it will generate that image based on your text command (like this article’s “strawberry frog,” pictured).

So ask your computer to draw, let’s say, “a single shaft of light illuminating a room with a canvas, on which there is a painting of a raccoon in the style of Rembrandt.” And the computer creates… this…

The ability of the system to parse and understand your plain language is a tell-tale sign of how AI will enrich human-machine interactions. Scan the gallery of images created by Imagen based on users’ wild, specific and often hilarious inputs (“a dragon fruit with a black belt in karate in the snow.”) and the amazing results is a fun (and impressive) exercise. Heck, AI is even designing magazine covers now!

But that begs the question: can you try for yourself? Sure. While Imagen is still a by-invite-only beta project, similar models – such as this one, based on the DALL-E Open AI project – are free to try. So enter any prompt you can imagine, and – in about 60 seconds – see how the computer draws what you’re envisioning!

And while you’re having fun with the above, note that as AI’s disparate capabilities become more and more mature, the idea of machines as idea-drivers… not just tools… is  another small glimpse through the window of what tomorrow will be.