Designing for a Human Experience in the Age of AI

These sacred UX principles might hold AI interfaces back. In fact, the best AI experiences often deliberately slow users down. Everything we learned in our careers turns out to be different from how we should actually work with AI.

Think about asking a friend for directions versus following a GPS. With your friend, those little pauses and just-to-confirm moments aren't friction. They're trust-building exchanges. 

Your friend might say, "So you're heading to the hipster coffee place downtown?" before giving directions. These natural checking points make you more confident in the guidance—not less.

This same principle should guide how we design AI interfaces.

Trust in AI can be earned through multi-prompt exchanges.

This insight hit home during my work launching an AI therapy service.

My initial instinct was to follow traditional UX wisdom: streamline everything with fewer steps and a shorter Time to Value (TTV). But I learned something surprising. 

People trusted the AI therapist more when it took moments to reflect and clarify, using phrases like: 

These seemingly inefficient pauses created space for understanding and validation. With this, we pivoted from suggesting recommendations after one exchange to multiple exchanges and delivering the right recommendations. Our early testers of the app report an average of a 22% improvement in their anxiety and an 18% improvement in their stress with each session.

People are having genuine conversations with AI, rather than issuing commands. 

A diary study of generative AI users revealed that half of all conversations led to follow-up actions, with approximately 53% involved multiple exchanges. Users weren't simply seeking quick answers. They were engaging in exploratory dialogues to build understanding. This pattern suggests that effective AI interaction isn't about commanding a tool. It's about fostering collaboration.

Additionally, users weren't just selecting options. They were having conversations, expressing uncertainties, and clarifying their thoughts. This makes the interaction messier but paradoxically more natural.

💡 Main takeaway: AI has a peculiar way of seeing the world. It thinks in patterns and probabilities, not buttons and menus

Collaborating with AI produces more creative results than surface commands.

This shift from commanding to collaborating transforms how users engage with AI. 

Consider two approaches to the same task. A user who simply demands, "Make me a bakery website," gets fundamentally different results than one who says, "I'm building a bakery website with a warm, rustic feel, and I need an ordering button that fits this aesthetic." 

The second approach invites the AI to engage with the context and purpose, leading to more nuanced and creative solutions.

Need even more proof? 

Some curious architects put this collaborative model to the test by creating 6,000 prompts and images using MidJourney, mixing both AI-solo creations and AI-human collaborations. They invited five experts for a design critique to judge the results.

Surprisingly, the AI-human pieces weren't just more creative. They stuck closer to the design brief. While the AI-solo creations got caught up in utility and function, the human-AI duets brought that extra spark of imagination.

The Big Takeaway

The future of AI interaction isn't about making everything faster and simpler. It's about making it more human. By embracing these natural rhythms of conversation and collaboration, we can design AI experiences that feel less like using a tool and more like working with a thoughtful colleague who asks good questions. 

This might mean more steps and slower interactions, but it creates deeper engagement and better outcomes. Here are some key principles to keep in mind:

1. Friction isn't your enemy. It's your secret weapon. Like that friend who makes you explain your terrible dating choices, the right kind of resistance leads to better decisions. AI should make you think, not just click.

2. Let users take the wheel. Sure, you could automate every choice, but then you'd miss gems like "Actually, I meant vintage-industrial, not industrial-vintage." Sometimes the scenic route gets you to better places than the highway.

3. Messy conversations beat clean workflows. Real breakthroughs happen in the "wait, what if we..." moments, not in perfectly optimized button clicks. Embrace the chaos; it's where the magic happens.

We are in a new and exciting time for design. Modern designers must embrace AI as a collaborative partner that challenges, inspires, and amplifies human thinking. 

The future of AI interfaces isn’t necessarily about efficiency, but about deeper conversations that transform both humans and machines.

About the Author

Co-founder of a venture studio called Studio.init(), where I partner with startups by trading design, development, and marketing expertise for equity. After honing my craft at Google, Coinbase, and Milkroad, I combine my HCI background and creative training to help people design and adapt to new technologies through psychology-driven methods. Follow me on LinkedIn or email me at [email protected]!

🚀 Boost your business with us—advertise where 10M+ AI leaders engage

🌟 Sign up for the first AI Hub in the world.

📲 Our Socials