A few weeks ago, I attended a course, exploring how user experience (UX) design for AI tools can evolve beyond the surface-level. It was equal parts energising and frustrating. Energising, because it confirmed that we’re at a genuinely pivotal moment for AI product design. Frustrating, because some of the most valuable ideas felt underexplored or sidestepped entirely.
We’re past the point where AI feels novel. But we’re also not yet at maturity. We’re in the murky middle. That means UX professionals, designers and strategists have a rare opportunity to shape the future of AI user experience before it calcifies into convention.
Bridging the gap in AI UX design: From exploration to application
One clear takeaway: while back-end capabilities in AI have advanced rapidly, the interface layer feels like it’s playing catch-up; most tools offer only rudimentary input/output relationships, with little consideration for new UX opportunities. AI is enabling ground-breaking applications, but to succeed, AI products must be built with clear, human-centred design that clearly meets user needs and expectations.
So why is UX so different with AI?
- Users don’t have clear mental models – leading to incorrect expectations of how AI-powered tools behave.
- AI systems are probabilistic -They can give different answers to the same question. That’s a design challenge when we have high-stakes interactions.
- Trust is fragile – Because outcomes vary, users often treat AI tools as unreliable or gimmicky unless trust is slowly earned.
Understanding these differences was a crucial insight from the course. If users lack the correct mental models, interfaces must explicitly guide them toward clearer expectations. As outcomes of AI-driven systems are probabilistic, UX needs to clearly communicate levels of confidence, helping users calibrate their trust appropriately. Building trust becomes not just about transparency, but about continuously aligning the interface design to evolving user expectations and habits.
Navigating the fine line between trust and innovation
When it came to strategic thinking about UX opportunities, I was surprised by how conservative the course was. Beyond the usual discussions about balancing utility and discoverability (like addressing the “blank canvas” problem in tools like ChatGPT), there was little exploration of what bold AI UX could look like.
The dominant recommendation? Be conservative. Don’t overpromise. Wait for maturity.
Being cautious makes sense – to a degree. Overconfidence can damage user trust, especially when bold initiatives aren’t adequately developed. High-profile experiments, like the premature launch of Rabbit R1’s “Large Action Model,” highlight real risks. Yet, caution shouldn’t paralyse innovation. Instead, it should encourage us to carefully choose which steps we take.
What if we were strategically bold?
There are still smart ways to take bold bets. One that stands out as being underexplored: AI systems that learn about us.
Why aren’t more AI tools personalising their responses over time? Why don’t they build a nuanced, evolving picture of who we are, what we value and how we work?
Take Spotify. It’s not perfect, but its personalisation strategy is quietly brilliant:
- It learns passively from how we interact (skips, likes, replays).
- It curates content in ways that feel both surprising and personal by using AI to analyse music and other media – making connections that traditional algorithms would be unable to.
- The system improves based on the collective interactions of the user base.
The result? An experience that adapts without intruding. Spotify adds value without demanding attention, building trust through consistency and subtlety rather than explicit prompts or interventions.
Imagine if AI assistants, research tools or even content creation tools did the same. Instead of starting from zero every time, they could build on past interactions, adapting tone, format and preferences. With AI, UX doesn’t have to be perfect out of the box – just visibly improving over time. The outcome? Users will be more loyal and find more value in your tools.
And personalisation is just one frontier. Others include adaptive interfaces that respond to user context in real-time, emotionally intelligent systems that adjust tone based on sentiment, or multi-modal tools that fluidly blend voice, visuals and gesture. And this doesn’t even consider the opportunities for new product categories enabled by AI.
So why isn’t this happening?
There are real barriers; privacy concerns loom large and short-term thinking favours quick wins over long-term relationship building.
But there’s also a lack of imagination. Instead of asking “how can we add an AI feature to this product?”, we should be asking:
- What experiences do users currently avoid because they’re too cumbersome?
- And how could AI simplify those tasks until they become effortless?
These aren’t just design questions – they’re strategic ones. Because once we shift from tweaking existing interfaces to rethinking interactions entirely, we open the door to new product categories, new value propositions and new relationships with technology.
The most transformative AI experiences won’t come from layering intelligence onto old paradigms. They’ll emerge when we challenge the paradigm itself.
Designing the next leap in AI UX design and human-AI interaction
AI’s potential will only be fully realised when UX professionals embrace their role as pioneers – not simply refining interactions but fundamentally shaping how people experience AI technology. By thoughtfully designing interactions that clarify, reassure and inspire trust, UX can bridge the gap between sophisticated backend capabilities and user-friendly interfaces.
This transformation isn’t about a single solution or technology. It’s about a mindset shift – moving from reactive design to proactively guiding users through uncertainty and complexity. And the teams that act boldly (and thoughtfully) will set the UX standards for everyone else.