Taste gets all the attention right now, and Anton Sten agrees it matters. His point, the one I can’t shake, is that taste doesn’t cover all of it. Trust sits next to taste, not inside it. A product can be tasteful and still make you hesitate before you hand it something that matters. It can be plain and forgettable and still earn the benefit of the doubt, because nothing in it makes you doubt.
His best line is about what happens when you ask a model to make something feel trustworthy: it’ll add a lock icon to the header and call it done. It reaches for the symbol because it can’t produce the substance.
I’ve spent years designing for industrial operators in critical infrastructure, and that line made me laugh because it’s exactly what bad security UX looks like: shields, padlocks, and green checkmarks stacked on top of a screen that hedges.
Picture an operator at a plant when an alarm fires. They don’t trust your interface because it has a security badge. They trust it because:
- it tells them what happened in the language they use on the floor
- it’s confident about what it knows and honest about what it doesn’t
- it never makes them stop and wonder what the system is doing
One flicker of doubt and they go back to what they trusted before: the phone, the paper procedure, the colleague down the hall. You rarely get that trust back in the same shift.
Anton says trust “shows up in how information is framed, the confidence of the language, the small moments where the product tells you it’s got this before you’ve had to ask.” Every item on that list is a research output. You can’t frame information the way an operator thinks if you’ve never heard one think out loud, and you can’t calibrate the confidence of your language if you don’t know what they’re afraid of.
Also in September: Taste, Thought, and Trust Don’t Come From the Model · Taste Is Judgment, Not Preference · The Four Cs Before the Prompt
Daniel Arevalo is a Product Architect and strategic product design leader. He helps product teams turn AI capability into trusted, adoptable workflows, with a background in industrial AI, IT/OT cybersecurity, and high-stakes enterprise software for critical infrastructure.