PHOTO GAME CASE STUDY
Online shopping can be a less than ideal experience. With shoppers needing to work within the constructs of software applications that rely on relational databases, finding what they’re looking for without lots of trail and error is difficult.
That’s because shopping based on information organized by text, metadata and tags is inefficient, disjointed, culturally based and cumbersome.
To take the next step forward, online retailers need to offer a more intuitive, far more visual online shopping experience.
The big question
Using deep learning algorithms, can humans train AI models to recognize photos?
Yes, they can, and here’s how Sentient Labs proved it.
We developed an iPhone game called Pixstachio. The game presented users with a target image, then served up sets of images, asking users to click on the image set most similar to the original visual.
Without knowing that Sentient AI was serving up the images sets based on their previous selections, users got closer to finding the target image with each selection they made, usually finding it within just seven taps.
In addition to being a fun game, Pixstachio proved that a more intuitive, more visual shopping experience can be achieved online without the need for indexed data or a text-based search, thus setting the table for retailers to transform the online shopping experience for their customers.
SOURCE: 2015 IBM STUDY: "SHOPPER DISRUPTED: RETAILING THROUGH THE NOISE"