AI Product Design: Flickr to Google Photo

Google Product VP on AI Product Design (Part 2)

After selling Virage and starting a new career path, I came across a new product called Flickr. Simply put, the product design blew my mind. Unlike our first Virage photo product, they figured out how to take computer vision to the next level through text-based search. For example, you could now type “sunset” into a query and receive millions of related images. Furthermore, these images shared a deeper level of shared attributes that went far beyond simple colors or shapes.

How did Flickr take image search to the next level? The answer is tagging. Instead of relying on the platform to categorize images, Flickr enabled users to tag a photo with a few keystrokes. There’s an incredible group of people who are obsessed with categorizing content – and they’re not paid to do it. Simply put, it’s a dedicated community of users that are out for the common good to make content easier to find.

This example illustrates that algorithms and AI only get you so far in product design. Ultimately, you have to talk to people and introduce humanity into the development process. I took these lessons learned from Flickr’s model and applied them to products at Google. Most notably, the product that became Google Photo was built upon early lessons from Virage and the social community aspects that Flickr pioneered.

We set out to create the “Gmail” for photos – building a mobile-first product that provides unlimited storage. In other words, we built the product around the camera that most people had with them. Today, this represents billions of people with smartphones as opposed to a handful of people with $10,000 cameras. As a result, we completely re-imagined the user experience to focus on people taking pictures with their phones.

In addition, we wanted to introduce a new organization method for photos. For example, you take a bunch of pictures on vacation – but then you have to sort through them and finding them becomes tedious. Using AI algorithms, photos are sortable based on similarities in scenery/location/people to make it easier for users to locate their favorite photos.

Ultimately, we figured out how to make AI do all the hard work for you. But, we also used lessons from Flickr to bring the human element into product design in order to engage with more customers.

Looking for Part 1? Click here to view.

Looking for Part 3? Click here to view.