Our thoughts on Dall-E: the text to image wizard
Open AI, an ‘Artificial General Intelligence’ company backed by Elon Musk & Microsoft, continues to push the boundaries of things machine learning models can do. After beating humans in a complex collaborative game (DOTA2) and writing an endless stream of comprehensible text with GPT3, they now focused on the creation of images from nearly any given text description.
For all of you who want to stay up to date on the wonders of AI-land, say hi to Dall-E.
Basically, the idea behind Dall-E comes down to this: you type in any sentence you want, and Dall-E will generate the image. We must admit, it’s an impressive technical achievement. Perhaps the use of Dall-E remains somewhat ‘gimmicky’ today, but that doesn’t mean it doesn’t contain real business value in the near future. Anyways, it triggered us to let our imagination run wild and come up with the following example use cases.
AI Pictionary
As easy and fun as it sounds. Someone enters some random words. Dall-E generates an image, based on those words. Send the image to a group of persons and let them guess the original words. Points for the one who finds them first!
Avatar generation
Your profile picture is outdated or you’re in desperate need of your own personalized avatar? Let Dall-E do the work for you. Just describe how you look (or how you want to look) et voila.
Illustrate a story for a children’s book
Children always love a good bedtime story. Parents with a rich fantasy might invent their own children’s stories from time to time. Wouldn’t it be great if they could write this story down and have Dall-E automatically provide the accompanying illustrations? Because a story without pictures is only half a story.
Virtual Reality app
Imagine a VR app where GPT3, the automated text generator, writes a narrative and Dall-E creates images to illustrate the story. Add some more machine learning technology for video frame interpolation and you can imagine a novel experience, where the content is fully created by models. You could even take it one step further by making it interactive. Pointing towards a particular item, for example, enables the user to steer the narrative and create a unique kind of experience.
Our thoughts
Admittedly, it isn’t easy to imagine how we’re going to use this in our daily or business life. But in many ways, it is exciting to think about the possibilities of this piece of technology. One way or another, it’s a new technological AI milestone. A new pathway for the usage of AI to add business value.
We can only suggest trying it out yourself on the original blog post. Make sure to let us know what you think of it.