Over the last decade or so, a trend has emerged where many junior designers (and even some fairly senior ones) “would not even pick up the mouse” without requirements — presumably because they got slapped on the wrist in the past for operating without complete approval of the leadership. 

I predict that this trend will undergo a major change in the next five years due to AI's emerging dominance. The design of AI-driven products is challenging because these products are not programmed – they are trained. They demand a new (old?) set of core skills and an experimentation-driven workflow (which I’ve written about here: https://www.uxforai.com/p/the-new-ai-inclusive-ux-process) in the wild land of discovery where the market moves at hyper-speed, and many existing patterns or clever things just do not work.

These core UX skills are not new – they are, in fact, quite old.

However, many of today’s designers need to re-learn these core skills of rapid experimentation, creating original ideas, conducting intense user research, and driving agreement through being of service to the team. 

Important Disclaimer

Someone Hands You an AI Project. Now What?

If someone gives you a chance to participate in a strategic AI project, they are not assigning you a new chore! They are going to bat for you and giving you a chance to shine. Make that project your first and only priority. It might be your one and only opportunity to shine at your current organization. Screw this up at your peril. ​​

When someone assigns you something, ask: “When do you need this by?” Write it down. Put it on a sticky note and stick it on your monitor. Get it done a few days early. Do NOT, under any circumstances, be late.

Your team is super busy and under tremendous pressure. Be a good corporate citizen: do not set up meetings with generic names or without agenda or goals. If you have lengthy questions or discussions, write them down in a document and share them. If someone asks you for a meeting on Thursday, do not propose it on Wednesday. Make sure it’s in the time that people did not have marked as “busy.” NEVER decline meetings without offering alternative meeting times. 

Be humble. Being a designer on an AI project is a privilege that is earned. During the meeting, offer to take notes. Learn how to quickly sketch other people’s ideas live during the meeting using a document camera. Show your sketches to the rest of your team right then and there. Add your sketches to FigJam boards and meeting notes. Be the one who RELIABLY and HUMBLY shares meeting notes and recordings with the team.

If you are not sure what you need to do, ask questions. Don’t just sit there like a complete noob twiddling your thumbs or, worse, complaining that no one tells you the requirements! Figure out WHO to ask. Figuring out WHO and WHAT to ask is part of your job. HINT: Your job is NOT to over-engineer your next nested auto-layout Figma monsterpiece. 

Speaking of Figma mockups, that’s all about to be done by machines anyway, that is – DIRECTLY CODED IN REACT, bypassing pictures (e.g., Figma) entirely. Repeat after me: “Figma is just pictures.” Code talks and pictures walk. Your job is to help your team get to the code for a new AI product faster and with fewer use case mistakes and usability challenges. Anything that interferes with that task must go, including your Figma auto-layout over-achievements.

When you create your prototype, be realistic about what can be accomplished in the assigned time frame. If at all possible, take care to use existing components. Make sure your new stuff looks and behaves the way your company’s old stuff already does. If you are not sure how it should behave, ASK. Asking questions is much more important than perfecting your next nested Figma auto-layout.

Figuring out which components are real and can be used in your project and which are “aspirational bullshit” of your fellow Imagineers is your JOB. The AI project might be your Golden Ticket to the party, but you are not living in the wonky “world of pure imagination.” You are not Willy Wonka, the Millionaire Chocolatier. You are not making everlasting gobstoppers. You are not making “high art.” You are making simple, practical solutions that work for a particular customer in this particular moment in time. In short, you are a designer, and you need to know what your chocolate cake ingredients look like. So dig deep into your bag of interviewing skills and go ask some more informed people what components you should use. Always check what’s already released and what’s on the truck. NEVER use future or imaginary system components unless SPECIFICALLY told to do so.

Stop complaining. Nobody likes a Veruca, the “Bad Nut” who screams when she does not get the golden goose component, like RIGHT NOW! (2005 Willy Wonka “bad nut” Veruca scene:  https://youtu.be/pt9tFXy0rdg?si=VvGR-c5pnfrbOnR4 or for classic’s lovers here’s the “bad egg” Veruca from the 1971 Willy Wonka: https://youtu.be/Pqsy7V0wphI?si=GZN5N1SymwtW9Lok )

Bottom line: you are a professional, so act like it.

Take care of documenting the basics. Every table should have a sort order. Period. No excuses. Describe the pagination (if any), search, filters, etc. Every object should have a CRUD. If you proudly deliver another nested Figma auto-layout monsterpiece that sings and bakes cookies but forget the basics like the table sort order, just what the hell are you doing?

If you do not like your company’s design patterns, or you just dislike popups in general, or you think re-using existing colors or fonts are beneath your delicate sensibilities – I have a suggestion: complain to the proper authorities! Tell your parents how much you hate it here. Or better yet, do us all a favor: tell your CEO that your boss or mentor is an incompetent fool who is making you COMPROMISE your UX PRINCIPLES to create the next chapter of the company’s future!

Do not, I repeat – DO NOT wait until the last minute to finish something important or visionary. Allow ample time for feedback from at least your primary and secondary stakeholders. That means a minimum of 24 hours of time per person. Double it if the time change is involved. If someone doesn't respond, query them again. STAKEHOLDERS OWE YOU NOTHING. If you are late, it’s all on you.

Do not simply send stakeholders your 50-page Figma file or a broken prototype to review. Send them actual screens with annotation arrows. Or better yet, shoot a ONE-MINUTE video with a realistic SCENARIO voiceover (do not exceed 1 minute; you’ve been warned.) Always ask specific questions about aspects of the design. Never ask, “Do you like it?” (Watch the brilliant Mike Monteiro explain why: https://youtu.be/niLVvoxhs5g?si=77NNR-0kOdHOy8Ix)

Articulating design problems is hard. Most of the time, NO ONE will be giving you a Jira ticket on an AI project. Why? Because if they could articulate what they wanted in a Jira ticket, they could just as easily ask the AI to create those fireframes. And AI will give them 5 or even 10 versions in seconds… Free. And without THE DRAMA. Show them why they need a human designer: add some value. Make your own Jira ticket. Write down exactly what you are going to do, when you will finish it by, and what the customer and corporate benefits will be. Check with your PM and developers for accuracy. 

Take the time to write down complete use cases in an agile format: “As a user X, I can do Y, so I get benefit Z.” Do NOT deviate from this principle. If no one gives you the use case, make one from the requirements or research it by reading your competitors’ marketing, training videos, or documentation. Send it for feedback ASAP, then iterate. Get it right. Show some initiative. Hustle.

Stop pulling bullshit prototype content out of your ass. It’s lazy and unprofessional. It invites everyone to treat UX like a fifth leg on a horse. (Google sometime what that means in Portuguese…) Learn what other people have been doing with their prototypes and match the content to the existing stories as best you can. I say that again — match the stories your team is already working on. Take the time to identify key points of content that might cause friction to users and create multiple versions so that you can effectively test it with customers and stakeholders. Create realistic, reasonable, and recent content for dates, reasonable data for titles, IDs, and descriptions that match actual content from real customers. ChatGPT will create great content for you and explain technical stuff in detail. 

Anytime you are not sure how to do something, research it. Ask ChatGPT. Look at the competition. Ask your co-workers.

Create prototypes with realistic content and with a specific purpose: 1) to demonstrate specifications to developers OR 2) for user testing. If you are adding animations and other slick Figma tricks to your AI-driven product prototype and they are not serving either purpose, you are just wasting cycles (and maybe, just maybe… <GASP!>… Showing off?) Fancy prototypes are a pain to create, maintain, and test, and speed of experimentation and direct user research are the keys to success on AI-driven design projects. Remember — “Figma is just pictures.” Code talks and pictures walk. Nine times out of ten, you don’t need the fancy prototype; you just need to talk directly with your team and your users.

Routinely produce multiple versions of the solution ON PAPER FIRST and ask people which they prefer and why. Evaluate solutions behaviourally – what specific actions are driving the customers to choose A over B?  If you can’t answer that, do more research. If you have the luxury of 2-3 days OR MORE to finish something, test it with customers. Can’t get to customers? Test with peers, developers, accountants, janitors, coffee shop weirdos, friends, or beggars. If you have not tested anything with other people, how do you even call yourself a designer? 

I’m serious –

You can pull complete UX for AI solutions out of your perfect ass, and they smell like god-damn daisies and tulips. Congratulations. You are fired.

If someone more senior than you gave you their source file, for God’s sake, take the time to learn how it’s been put together. Don’t start changing components unless someone SPECIFICALLY asked you to. Only ever change what someone asked you to change. And for god’s sake, document WHAT you are changing. Add comments on wireframes and make a summary sheet describing what you changed in each of your versions. That is also more your job than all of your over-engineered nested Figma… Well, you get the idea.

One last thing: if someone tells you something is urgent, and the boss wants it right away, just do it. Burn your weekend, pull an all-nighter – whatever. Get it done. If someone abuses this “emergency release lever” in the future, change your behavior THEN. But the first time someone asks you, just get with the program and get it done.

Got it? Now go and do it.

In Conclusion

I have been a designer, a leader, and a mentor of designers for over twenty years. I LOVE designers. However, many of today’s designers are poorly equipped for the new age being ushered by functional AI.

That is why I am concerned about the future of our great experiment.

The good news is that CORE UX Design skills, even rusty ones, can be of tremendous value in the new normal. More than bringing value, Designers are essential in bringing “Balance to the Force” because AI is simply too important to be left to only data scientists and business people. 

Thus, I hope this letter inspires you in all the right ways, instills an occasional chuckle, and gives you a sense of urgency to rethink and retool. 

Self-starter action and original ideas, together with driving broad project alignment, coordination of conflicting interests, and lightweight design process centered on user research – these skills will now be more valuable than ever. Due to the major disruption introduced by AI, these skills are once again becoming the bread and butter of our industry.

Designers! Step up – The World needs you!

With love,

Greg — my opinions are my own.

P.S. Interested in learning more? Check out my upcoming Full-Day workshop at UXSTRAT: https://strat.events/usa/tickets (Be sure to get the “pre-conference workshops,” still available at a discounted early bird price.)