You must have seen the sparkles either as a button or beside the words ‘Enable AI’. Those glittering stars that have become the universal symbol of AI-powered. Like the ‘Intel inside’ of today. They sit next to text inputs, lurk over buttons and fall all over interfaces like it’s a star storm in Animal Crossing. However, what many are adding to their site thinking is a sensible way to indicate, to guide is now being also used as a warning label by some. A way to know not to click, not interact and not to trust what comes next from that interaction.
It all started sensibly. It feels right that having a universal indicator for ‘AI’ on paper is a good thing. However, what those sparkles have become to many is an almost Pavlovian response to scroll past, click away and avoid the artificial of what comes next. It’s not the sparkles fault, it is the fault of just sticking a sparkle on it and thinking that is a job well done.
When guidance becomes gatekeeping
The sparkle icon was meant to solve a real problem. Users couldn’t tell what features were AI or not. The sparkles served as a way finder, a gentle approach to say ‘hey this is different and curious to click’. The thing is humans are quick to attach baggage to iconography. That one time you click a sparkle and it doesn’t give you good results, that other time it was clearly hallucinating. That means you are ruling on all sparkles.
This is an issue when a symbol for interaction is adopted universally across different interfaces. Another interface letting someone down can even if you have the best interface domino effect yours and the person coming to your experience is tainted already. Their sparkles aren’t shiny, they are expecting to be broken and flawed.
The conscious avoidance
In the past few weeks I have heard quite a few people refer to the ‘sparkles’ as an easy way to know what to avoid. Often they do it in half joking, the half that isn’t joking is the bit that no matter how useful the experience, the moment you bring a sparkle to it you’re negating everything. People have developed an intentional resistance to AI-branded features. If we think about hype this is natural and a part of the ebb and flow of technologies.
Take a moment and think about your own behaviour. When you see a sparkle, what do you think? Are you drawn to engage? Do you look for them on interfaces as entry points. If you think about it for a moment they seem a little ‘dated’ now. Often littering the interface. The moment you start looking for sparkles they are everywhere, it’s like an infestation.
This is the paradox we have created; the more something is signalling AI-powered, the less likely people are becoming to use it. The bind here is that this is the case even if the implementation is useful. Another way needs to be found. Perhaps one that is less sparkly.
Beyond sparkles
Whilst putting a sparkle on it is a predictable pattern. If this is falling short, what is the option? To me, this is where the next and more exciting stage of design comes in and that’s integrated features for AI. We have today a step towards that with what maybe you could label the ‘click and drawer’ approach. AI features collected, not sprinkled everywhere into one useful spot. I would argue this is often clumsy and shoves AI out of mind.
The most successful AI integrations are the ones you rarely notice. The ones that don’t announce themselves with visual cues and make tasks easier. That help you get the job done and build trust, don’t shout about how amazing they are going to be at it. They are embedded into the experience, not signposted everywhere or hidden.
Bringing to the experience
The solution I would point out isn’t to remove AI capabilities or hide them in a drawer, but to embed and weave them into the workflows. To make them as native as every other piece of the experience.
“The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.”
Mark Weiser
Instead of cosmic iconography or special menus, drawers that sparkle and hide, embedding where users already work is the way forward. AI should feel like an enhancement of existing capabilities, not a replacement that requires learning new interaction patterns.
If you consider email writing. Rather than adding a sparkle ‘writer’ button that opens a new interface, why not improve the spellcheck or suggestions. Instead of a chat box widget, why not enhance the search to understand context and intent more naturally?
This is a focus on making existing workflow genuinely better for the people already using products. This is how you stop people ignoring the sparkles and using AI, embracing and seeing how it can bring them success in their day-to-day. We then stop making AI feel like a separate, too magical thing. We stop giving reasons to avoid it and build trust.
AI isn’t useful as a sparkle because it’s not magic and will never live up to that. It’s useful when it solves a real world problem.