Trending Articles
Trending Topics
Trending
This article is part of our Opinions section, where we invite industry professionals to share their views on the most pressing technology questions of our time.
Is AI really the first genuinely new UI paradigm in decades? There are some big bets on that it will be: see OpenAI’s recent $6.4bn acquisition of an AI device startup founded by the guy who designed some of Apple’s most iconic products, including the iPhone and iPad.
Ive and Altman have gone on record about wanting to explore the next major shift beyond tapping and swiping: computers that feel less like tools and more like collaborators. It’s tempting to see this as a repeat of 2007, when Nokia executives dismissed the iPhone as a toy that “couldn’t even make proper phone calls”. Months later, the App Store arrived. Touch interfaces transformed daily life.
AI-driven interaction stands at a similar inflexion point. But focusing on the exact input mode for interaction feels like the wrong bet. If AI assistants and applications are truly going to win consumers over, it won’t happen because of surface-level design aesthetics. AI assistants have to evolve beyond the design of user-prompted interfaces, into anticipatory support that exists on demand, at the right time. It will require a ‘social design’ that fully leverages user context and history to create the ultimate personalised experience for individuals.
Unlike the iPhone moment, this transition probably won’t be nearly as quick or smooth.
UI design has, for decades now, meant guiding people through layers of menus, buttons, and screens. Booking a restaurant involves multiple taps, searches, and confirmations. The promise of AI is a reduction of that friction: the freedom of simply vocalising, “book the same Italian place for Thursday at 7pm”. The system uses history, context, and integrations to handle the task end-to-end.
Of course, people like to focus on voice interaction as a marker for AI transformation because it’s a visible element with some novelty to it (and maybe we’ve all just watched a bit too much sci-fi). I sometimes hear people asking whether voice will be ‘the next UI’ that dominates AI interactions. Humane AI Pin and Rabbit r1 are just a couple of companies that have already tried to replace the classic form factor of a smartphone with voice.
There’s at least one good reason why these products haven’t taken off. It’s the same reason why voice won’t be the ‘one UI to rule them all.’ Boiling the evolution of AI down to ‘replace clicks with conversations’ misses the point of what we need from AI. The more important UX shift for AI is something far more invisible: not the buttons we press, but the intelligence behind them. AI-driven interactions need to be built around anticipatory social design. AI assistants must transform from user-initiated, reactive input to proactive support. Imagine: “Based on your calendar, should I move your meeting, since traffic is unusually heavy?” or “Your friend is in town. Should I suggest dinner spots you both frequent?”
That’s where AI will have a real impact on UI/UX. It won’t be in the front-end, but in the non-visual stuff happening behind the scenes – building a long-term understanding of you, learning your preferences. Like becoming a friend in real life, this takes a long time to do.
Even if AI assistants get really good at understanding users, there are some other pain points that AI will need to overcome before consumers accept the technology as here to stay. The biggest is reliability.
AI assistants might already be able to maintain context across multiple exchanges and combine conversation with visuals, but they struggle with latency. Interfaces will have to support low-bandwidth scenarios or older devices, and ensure that both voice and text work offline or at least with minimal latency. Internet infrastructure, to this day, is far from perfect, and not everyone has premium connectivity. As long as that remains the case, no company can rely 100% on the cloud. Similarly, conversational AI cannot be a feature that only works flawlessly on the latest flagship phone. Unless AI works ubiquitously for everyone, AI will never move beyond being a luxury item for the few.
The automotive industry learned this lesson the hard way. Car manufacturers once chased ultrarealistic dashboards powered by game engines, only to find they couldn’t deliver those experiences on production hardware at scale. The lesson was to design for practical reliability across performance levels, not just fancy demos.
Anyone developing AI assistants will have to take on board that lesson. Part of this means understanding that each input has its own unique strengths and weaknesses – and crucially, that users have their own preferences for these modes. New modes won’t replace old ones; buttons, touch, and gestures aren’t disappearing. Why? Because sometimes, tapping a button is simply faster and less error-prone than speaking aloud.
That last point is especially important when you consider the cultural and language differences between users worldwide. We’re probably a long way off from the future hinted at by a Google I/O demo in 2022, which showed real-time translation via smart glasses of English to Mandarin, spoken aloud with natural intonation. Creating a future where communication barriers fall away and AI is seamless across devices and dialects will take a lot of time and innovation.
AI-powered UX will succeed when it solves real problems seamlessly: remembering preferences, anticipating needs, and removing friction without drawing attention to the complexity behind the curtain. The best interfaces should feel effortless and consistent.
Again, achieving ‘the best’ UI/UX will not necessarily mean chasing sci-fi trends of holographically projected AI assistants. We have to look beyond audiovisual, focusing on predictable behaviour, trustworthy data handling, and respect for context. With consumers valuing personalisation more than ever, this will be the dealbreaker.
I suspect progress will be uneven, with many blunders to come, with some industries leaping ahead while others remain in transition. As computing costs (hopefully) fall and AI moves to the edge, we could well be looking at fully personalised, real-time, context-aware interactions even on modest devices.
Yet, the most transformative shift will happen quietly: AI systems that truly understand users, anticipate needs, and act in ways that feel natural and trustworthy. When the UX of AI applications becomes less about talking to computers, and more about building relationships with technology that understands us – not only in what we say, but in what we need – that’s when AI will win everyone over.
