It’s very unlikely that web designers are going to be replaced by computers in the near, or far, future, but there are some interesting developments in the world of automation that could make your job easier, or at the very least, more interesting. One of the more ephemeral developments has come via Google’s “Dream” Code, which is based off a rather complex neural net that the company utilizes in its image search and recognition technology. By essentially telling the program to find an image of something specific in an image that doesn’t actually have that detail, the coding will actually enhance edges and “create” that image where it wasn’t originally featured.
How does this all actually come out in practice? If you’ve ever looked at the sky and seen a group of clouds, and then imagined what shapes those clouds might be, you could define certain parts to be an arm, or a leg, until you have the shape of a rabbit in your mind. With the case of Google’s Dream code, the image of a rabbit shape would be enhanced, and made very “real.” The code can find images of dogs in abstract galaxy shots taken by satellites, or turn a selfie into a surreal collection of neon eyes. Whatever the outcome, this amusing novelty is actually part of a larger, fascinating development in automation, and one that could have some serious implications for web design.
Recognition is the Key
One of the hurdles faced by artificial intelligence is the actual ability to recognize what it’s looking at. Google, and other companies involved in software and artificial intelligence, have been going to great lengths to create programs that have an easier time understanding what an image holds.
For web design, although we aren’t quite there yet, there may be some interesting tools and opportunities for developers. Imagine being able to create not only responsive, resizing websites, but to also create entirely unique responsive graphics based on choices that a user is making. For example, if your site sells certain gourmet foods, an intelligent image selector may “reimagine” the site’s graphical layout to put more emphasis on those foods in image selections, while maintaining your own parameters for color, shape, and more. In other words, the machine could act like the subconscious of your visitor.
The implications are even more interesting for advertising. If you can imagine ads that are unique to each visitor, based off of recognition of different information that your site has received about time of day, habit, and even time spent browsing or reading a particular article, it becomes easier to picture just how effective a “smart ad” might be. It’s just one of the ways in which a machine with image recognition could make a difference in web design. In time, sites could be entirely customizable based on actual user activity– a future which could lead to some fascinating design choices and plenty of work for the designers themselves.