Microsoft Created An AI That Is Able To Draw Something Original, Just Out Of Plain Text

Something from nothing. This is what to describe what Microsoft is doing with its AI.

The company has developed an AI to draw something that is entirely an original image, based on nothing more than plain text. What this means, using a simple query, the computer that uses the AI can draw something out of it. For example, typing a descriptive words like: "a bird is blue with white stripes has a long beak."

This is showing how humans are getting closer in making computers understand the world we live in.

To do this, the researchers at Microsoft use a neural network called an Attentional Generative Adversarial Network (AttnGAN), which is able to create images pixel-by-pixel. Then they were able to create a text-to-image bot that can draw pretty amazing pictures.

Just like how artists or designers create their masterpiece, the AI draws by broadening strokes and create fine details in layers.

According to a blog post by Xiaodong He, AI research manager with the group:

"If you go to Bing and you search for a bird, you get a bird picture. But here, the pictures are created by the computer, pixel by pixel, from scratch. These birds may not exist in the real world — they are just an aspect of our computer’s imagination of birds."

Ai in able to recreate a picture of birds after being shown pictures of birds, isn't particularly impressive. But what makes the Microsoft's project different, is that it can generalize and behave on inputs it has never been presented with before.

The AI is able to pick up the meanings of individual words in the sentences, rather than matching complete sentences to complete images.

This can be useful because computers, no matter how smart they are, have long been known to have no imagination like humans do, making them have no sense of inspiration. The researchers here, is like debunking that belief.

The potential usage of such technology, is in the design industry. This can open ways for designers to create images by just spelling the words out: telling their virtual assistants to draw something with a simple voice command.

The Deep Learning Group at Microsoft created the AI as part of a trilogy of AI projects. The projects were developed to provide useful applications which combine both computer vision and natural language processing.

The idea here, is to tech machines how to understand humans and the world.

Published: 
22/01/2018