Google is one of the leading companies in the tech industry, and that includes Artificial Intelligence (AI).
The company has launched a tool called AutoDraw, which is similar to Microsoft Paint but with a AI brain.
With a simple user-interface, users can draw doodles using a rudimentary pen tool via their mouse or touchpad. Then the app will provide a list of possible images that it thinks users were trying to create, Users can then choose one, and the app will turn the doodles scribbling into proper images.
The software is based on the technology that was introduced on Google’s previous doodling experiment, Quick, Draw!.
Like a game, using Quick, Draw!, users were asked to draw six simple objects in 20 seconds while its character-recognition AI attempted to identify them. But at the same time, the app also gathered information about the scribbles.
When the amount of data gathered are huge, Google then was able to create AutoDraw, a recommendation tool, not anymore a game.
While the interface and how the AI works seem simple, but the technology is what pushes computers to understand more complex and unreliable communications of humans.
Google isn’t alone in trying to solve this problem. There are others that are trying to develop ways for computers to understand drawings. For example, the AI firm Gamalon described how it was developing new ways to identify similar scribbling using less training data than traditional AI approaches.
AutoDraw itself can be a good tool to see how computers have evoived. However, the results are likely to be mixed. It can guess scribble of a spider and a frog, for example, but mistakenly interpret a bird as an Eiffel Tower.
AI technology in this field is far from perfect. But humans are sure headed towards the right direction.