There are four ways a product can express uncertainty and no fifth one.
Show confidence: give the answer, attach a signal about how sure the system is. Hedge: give the answer, soften the claim inside the words. Stay silent: show nothing rather than a weak answer. Ask: get the missing input from the user instead of guessing.
Almost every team chooses based on how uncertain the model is. Low confidence, add a badge. Wrong input. What matters is what a wrong answer costs the person reading it, and whether they can check it.
| If being wrong costs | And the user can verify | Use |
|---|---|---|
| Little, and it is easy to spot | Easily | Hedge |
| Little, but hard to spot | With effort | Show confidence, with the source |
| A lot, and it is recoverable | With effort | Ask |
| A lot, and it is not recoverable | Not really | Stay silent, route to the human path |
Teams get the bottom row wrong. When the mistake is expensive and the user cannot check the work, a confidence badge does not transfer the risk to them. It decorates a decision they are not equipped to make, and then everyone feels covered. Worse than saying nothing, because it manufactures the appearance of informed consent.
On showing confidence, the forms are not equal. A cited source opened in one action is the strongest signal there is, because it lets the user do the checking rather than trusting a number. Coarse bands work when they change behavior, and numeric scores are almost always worse than bands. The failure mode is furniture: a signal shown on every output, always in the same place, is invisible within a week.
On hedging, the copy that works is specific. "Based on the last thirty days" instead of "Your usage is." "This looks like a duplicate" instead of "This is a duplicate." Universal hedging fails: "I might be wrong, but" attached to everything is the same as no hedging.
Silence is the most underused pattern in the set.
It is also the one that separates products people trust from products people double-check. In practice it looks like the suggestion not appearing with the manual path right there, or a plain line about not having enough to go on yet plus the one action that would change that. The control stays available and returns nothing, because a control that vanishes reads as a broken product. And silence needs a sentence, which is a next step rather than an apology.
Asking has three rules: one question rather than a form, an account of the work already done before the question arrives, and an honest default. "Assume last quarter" is a good default. Pre-selecting the answer you were hoping for is not. Do not ask as a reflex, which moves the system's uncertainty onto the user's calendar.
Three things never to do. Never express uncertainty with visual weight alone, because greyed-out text is an accessibility problem rather than a confidence signal. Never let the uncertain state be less designed than the confident one, since the uncertain path is the common one in a generated product. Never apologize twice.
The artifact is a table with one row per surface: the cost of being wrong, whether the user can verify, the pattern, and the exact copy. That last column matters more than it looks. Uncertainty is expressed in words almost every time, and if design does not write those words they get written by whoever is closest to the code at eleven at night. Which is how one feature ends up apologizing in one place, hedging in another, and stating a guess as fact in the third.
One thing to do this week: take the surface where your product is most often uncertain and write the exact copy for all four patterns, in full sentences you would ship. Then pick one and delete the other three. The deletion is the design decision.
The four patterns with their failure modes, the selection table, and the blank library are in The Uncertainty Pattern Library. The full argument is The Shape of Uncertainty.