Google's Gemini team skipped cute mascot characters for flowing gradients and circles — does that actually build more trust in AI?

This explores whether abstract visuals like gradients build trust in AI differently from character mascots. The short answer: the collection has no head-to-head study that tests gradients against mascots with real users. What it has is one major design decision that rejected mascots on purpose, plus several trust studies that help explain why that choice might make sense, and where it might go wrong.

The decision comes from Google Design's work on Gemini. The team turned down both object icons and mascot characters. Instead they used circles, gradients, and motion to show AI 'thinking' as an energetic process rather than as a thing or a creature Can abstract gradients build trust in AI systems better than icons?. The reasoning was that gradients pull the eye and suggest momentum, and motion signals that something is happening inside. Note what the design does and doesn't do. It makes processing *feel* transparent without showing any of the actual mechanism. That is a design rationale, not a measured result.

The strongest case against mascots comes from research on anthropomorphism. Nielsen Norman Group describes four escalating levels at which users treat AI as human-like: courtesy, reinforcement, roleplay, and companionship. They name personality and emotional design as specific levers that push users toward misplaced trust, especially in task-oriented work where competence should matter more than charm How far can personality push users toward misplaced AI trust?. A mascot is essentially a personality layer you can see. Seen this way, choosing gradients is a way to avoid speeding up anthropomorphism, not just a matter of taste.

The less comfortable point is that abstract visuals don't escape the trust-heuristic problem. They change which heuristic users rely on. Users already trust ChatGPT because of its conversational responsiveness, independent of whether it's accurate Does conversational style actually make AI more trustworthy?. They also prefer answers with more citations even when the citations are irrelevant Do users trust citations more when there are simply more of them?. A gradient that 'looks like thinking' could work the same way: a signal of reliability that isn't connected to actual reliability. Research on visual rationales adds a caution. Graphical explanations helped people calibrate trust on verbal tasks but hurt calibration on visual ones, and how satisfied users felt didn't track how well calibrated they were Do visual rationales help or hurt how people calibrate trust?. Visuals that look trustworthy and visuals that help people judge correctly are different things.

If you want the alternative to both mascots and mood-setting gradients, look at trust scaffolding. These are UI patterns that let users check the AI's reasoning (citations, chain-of-thought, a source browser) and control its permissions, and that disclose failures plainly Do UI patterns for verifiability actually build user trust?. The real question underneath 'gradients vs. mascots' may be whether visual design should create a feeling of trust or give users ways to earn and check it.