AI-First Product Design

Disclaimer: This article was written without the help of generative AI.

The term “AI-First’’ came to my mind recently, when reviewing a design for a new feature our product team is working on. Without going into too much detail, It had something to do with protecting Personally identifiable information (PII). The team was thinking about providing users who encounter PII in the system with ways to protect that information.

The initial approach required the user to define rules, which often involved writing fairly complex expressions. We quickly started thinking about how we could use AI to help the user write those expressions. But as we were discussing this further, it became clearer that we needed to turn the entire thing on its head. Instead of letting users create these rules as they encounter PII, we should think about how we can use AI to have the system detect PII and apply protection rules to it automatically, without the need for the user to set any rules or write any expressions. This fundamental shift in our design approach was what made me think about AI-First. It reminded me of the early days of the smartphone revolution when one of the hottest topics in the UX community was Mobile First.

A little about Mobile First

Nowadays, mobile device displays are usually pretty large and have high screen resolutions. However, not so many years ago, mobile devices had relatively small screens with a low screen resolution. For example, the iPhone 4 had a screen size of 3.5 inches and a resolution of 960x640. This was when desktop displays were typically not smaller than 14’ and had screen resolutions of 1024x768 or higher.

That means there is not enough real estate on the mobile device screen to display all the elements available on the desktop version of the website.

Another aspect (pun not intended) is that most websites were designed for a landscape screen orientation. In contrast, although mobile phones have both landscape and portrait orientations, the “natural” orientation is portrait (by this, I mean that most users, at least by default, hold the phone vertically in their hand).

Also, the mobile networks were much slower than today, and every KB that needed to be sent over the network mattered. Load time is critical, especially on a mobile device, hence reducing page weight (KB) to the absolute minimum necessary was crucial. Even today, with 5G, when the network is much faster, response time remains critical, and page weight still matters.

At first, the common approach was to “force-fit” websites and apps designed for the desktop to the mobile phone screen. This required reorganizing the content and the layout of the website, tucking menus under hamburger menus, making elements smaller, and even removing certain elements and content altogether. This is often referred to as Graceful Degradation — designing for the large high-resolution displayed and adjusting the design for smaller displays with lower screen resolution.

Mobile First, Introduced by Luke Wroblewski in his book from 2011, is an approach to product and UX design in which you first design for mobile devices and only then design for the desktop. In his book, Luke calls for a fundamentally different approach to mobile design: Progressive Enhancement instead of Graceful Degradation. This means designing for the smaller screen first and then considering what should be the experience when moving to a larger display with higher resolutions.

Many think about Mobile-first primarily in the context of the form factor (i.e., fitting the content on a small(er) screen). However, this is missing some of the fundamental aspects of what it means. Mobile-first means thinking about the mobile experience first. It is not just about the text size, layout, graphics, etc. It is thinking about how a user is going to use the website or the app while on the go. It’s thinking about the user needs and how they are different when using the website on their mobile device vs. on the desktop. Location is probably the most obvious difference, but there are also other aspects like time of day, context, and even privacy that should also be considered.

“Location and time act as constraints on the mobile design process because they force you to think differently about how people will use your products throughout their day. They also create new opportunities for engagement that can help you innovate. So let’s talk about the new things mobile allows you to do.”

Mobile First, Luke Wroblewski

In short, mobile-first is not just an approach that helps optimize the mobile experience. It is a mindset you should have when thinking about the experience.

AI-First

The smartphone revolution was one of the most significant disruptions in the user experience design industry. It required rethinking the design and interaction from the ground up. It changed dramatically standards and patterns as well as user expectations.

With the recent rapid developments in AI, we are now facing maybe even greater disruption. AI is already changing the way we interact with digital systems, and in the coming years, as it continues to progress, we can expect even more significant changes.

Many of the existing products are trying to figure out ways to enhance and improve their product using AI. This often means taking existing features and functionalities and enhancing them with AI-empowered technologies. Some examples are automating existing tasks, providing summaries and insights, and improving the quality of the system responses to user requests.

In some cases, these are very useful and powerful improvements while in many other cases, we (still) see superficial implementations and sometimes only a veneer of AI on top of existing functionalities, mainly for marketing purposes (we need to be able to tell our customer that “we have AI in our product”).

As AI advances, the way we interact with technology will change drastically. We will see more and more of the tasks that users need to do today being automated. When user interaction will still be required, It will probably become more conversational. We would be able to just talk to the system, have it understand what we want, and perform complex tasks for us. The existing paradigm of point-and-click (or tap) interactions will be much less prevalent or maybe even fade away. The famous saying goes, “It is difficult to make predictions, especially about the future,” but I think it is safe to say there are at least a few years before this future becomes a reality. In the meantime, the basic ways we interact with digital systems will continue to rely on the existing paradigms.

The dramatic advancement in AI is the context in which I think we should start using an AI-First UX design approach, somewhat similar to the mobile-first approach. With an AI-First approach, instead of starting with an experience and then asking, “What can we do with AI to improve it” you start by thinking about how you could solve the problem differently with AI and only then craft the experience around it.

This is actually not a new concept. Google, for example, started talking about AI-First several years ago. Here is a quote from Google’s CEO Sundar Pichai in his Google I/O 2017 keynote:

“Computing is evolving again. We spoke last year about this important shift in computing from a mobile-first to an AI-First approach. … In an AI-First world, we are rethinking all our products and applying machine learning and AI to solve user problems.”

Sundar Pichai

Going back to the example of the PII project I started with, we went from thinking about how we could solve the problem using the “traditional” user experience, moved to think about how we could help the user and improve that experience, and ended up taking an AI-First approach to the problem. Cases like this can probably be found in almost any product. I suspect that in many cases, you might encounter pushback from a product manager who wants to deliver something fast and would say that using AI would take too long. Or engineering teams that would say it is going to be too difficult or expensive. However, the value provided to the user and the improved experience may be significant enough to justify it. Adopting a mobile-first approach required a change in the mindset of the entire product team. AI-First would require a similar change.