Become an AI Designer: Here's the Kind of UX You'll Probably Be Working on Soon

This is Part 3 of a 5-Part series on how AI is revolutionizing the field of UX design, and how to keep your skills sharp. Read time ~3 min. If you haven’t read Part 2, read that first!

In my last post, I shared how knowing the timeless fundamentals of our design discipline are still essential, if we want to direct AI tools to generate professional-grade results.

But given the sea-change that generative AI tools represent for pretty much every industry...

Not only will we be using generative AI tools and services, we will increasingly be making them.

Clients and customers will need design expertise to lay out AI-driven user flows. So what do we need to know to be great AI designers?

Design with AI for AI

As mentioned in Part 1, a big part of getting a solid output from AI tools depends on feeding them the right input: the art of promptcraft.

From generating an image to decorating a physical space, AI UX flows will guide the user through a prompt generation process.

When the user is still exploring approaches before committing, onboarding flows will coach them through prompt syntax.

There will be a lot of attention invested in auto-suggest experiences. Views that expose the capabilities of AI tools will help users know what's available, so they can see which variants align with their aim.

As users move past that initial exploration phase and start drilling in on a chosen direction, the workflow will focus more on refinement. Helping this along, a recent major game-changer is that tools like ChatGPT can now remember the context of previous prompts.

An interesting challenge for designers will be guiding users through iterative refinement of their prompts. We can help users pivot their outputs in the direction they want to see.

Early pioneers are also exploring ways of using AI to sculpt the prompts themselves. For example, you can ask AI tools like ChatGPT to generate higher-quality prompts for you to use in other apps like Midjourney. In other words, you can use prompts to get better prompts.

Speaking of Midjourney though...

An early challenge for AI designers will be addressing the elephant in the room. There's a tangle of pressing questions about AI art copyright, attribution, and plagiarism.

When so much of the output is trained on images that may not be with the creator’s consent, it raises red flags about IP.

Teasing out these tangles into systems that benefit & incentivize creators will soothe a huge pain point for artists, and present a big market opportunity.

Designing ways to credit the content an AI was trained on will be the first steps toward building equity for creators, and even paying them royalties.

For more on this subject, here’s my take on how AI will affect job scarcity for creatives. (Coming soon)

As designers, we also need a point-of-view on where AI does and doesn’t belong.

Companies are flooding into the space of ‘AI for (insert service)’ under the rallying cry of ‘Cognify ALL the things.’

But in the rush to do so, designers should pause to ask whether we’re shoehorning AI into our products without asking our users if that’s what they want.

The backlash against popular digital drawing app Clip Studio Paint’s announcement of AI features is a perfect example. (They walked it back and apologized. The people have spoken!)

The possibilities for AI are growing FAST, and that can make it hard to keep track of the basics. We’re all still humans, solving problems for other humans.

On that note, a very important job for UX designers will be thinking through how to integrate guardrails for AI tools to prevent psychological and societal harms...

Next post, we’ll take a look at our great responsibility to design safe, inclusive AI experiences for diverse audiences... and the intense high stakes of getting it wrong. We'll also examine some of the landmines we’ll need to overcome!

Go here for Part 4.