The Internet has changed dramatically over the past few decades. What was once a strange piece of high technology has become commonplace. In fact, an entire generation has now grown up never having lived in a world that wasn’t digitally connected. During that time it’s gone from a system that was little more than small personal pages to something that functions as both library and mall. But through all of the changes there’s been some things which have remained the same. But a recent article in the MIT Technology Review points out that this might be changing. One piece of legacy technology is finally making way for something new. The old is traditional text based search engines, and the new is AI based visual search engines.
The article brings up the fact that people typically browse by sense of sight. Few people choose items based solely on a list of features. Instead when people look for a variety of items, they tend to do it by contrasting and comparing based on sight. They find something they like the looks of and then compare it to other items with similar traits. This helps them to refine their idea of what they’re looking for. But search engines up to this point have been based more along the lines of a checklist.
But several companies are changing how this is done. There’s a few in particular which are ahead of the rest. But one of the most noteworthy is a company called Slyce. One of their biggest breakthroughs has come about through the use of distributed networking in relation to deep learning algorithms. Deep learning is a subset of the AI field. And the combination of AI and Visual Search has proven very successful. But it also requires extremely powerful hardware that’s beyond the scope of both desktop computers and smartphones. But Slyce has created a system of powerful servers within their own network.
Slyce’s servers wait for a signal from devices that make use of their API. This might come from a smartphone, or from the website of someone who has a catalog of products to sell that have visual components. Slyce’s powerful learning servers are then able to process the images and make decisions based on the results. For example, it might notice someone’s interest in boots that have a certain visual style to them and then offer suggests about similar boots or even outfits at the same store.