Confident wording hides uneven evidence quality and uncertainty.
← Back to patterns
The Design of Everyday Things People + AI Guidebook Information Foraging with Generative AI
Trust & Verification
Confidence Signals
Not All AI Confidence is Created Equal
AI communicates its certainty level per claim with color-coded badges, source counts, and expandable citations. Combats automation bias by making uncertainty visible.
Metaphor
Trust Meter
Interaction
Visual, Expandable
Use case
Research, Fact-Checking, Decision Support
Failure mode
Hidden State
Control mechanisms
Color-coded Badges · Expandable Sources · Dig Deeper
Reference examples
Consensus Claims & Evidence Table · Scite Smart Citations
Interactive Example
The live side-by-side is the comparison.
This prototype already contains the contrast between chat and the pattern response. It stands as the primary comparison surface for this page.
Fluent prose has one tone. The evidence underneath does not.
A chat answer renders a claim backed by forty studies and a claim backed by one shaky blog post in the exact same confident voice. The reader has no way to tell them apart, so they trust the weak claim as much as the strong one. Plain text flattens uneven evidence into a single flat register.
Confidence Signals attach a color-coded badge, a source count, and an expandable citation to each claim, so uncertainty becomes a visible property of the sentence rather than something buried in hedging language. Low-confidence outputs stop reading as authoritative and start inviting correction and deeper research.
The most dangerous AI claim is not the wrong one — it is the uncertain one that sounds certain.
DOPE Lens
How this pattern improves AI interaction quality.
Observability
Uncertainty is visible, not hidden behind confident-sounding text
Directability
Low-confidence outputs invite correction and deeper research
Predictability
Color coding creates consistent trust calibration
Explainability
Sources, methodology, and study counts are visible for every claim
Fit
Where it earns its place.
Use it
when
when
Answers mix high-evidence and low-evidence claims and the reader cannot tell them apart.
Avoid
when
when
Reliable confidence estimates do not exist and the signal would be decorative — or the reader has no lever to pull, where a confidence band transfers anxiety without transferring any control to act on it.
In the wild
Where this pattern already ships.
Real implementations from products users already know. Each entry shows how the pattern got applied — what shipped, where it sits in the product, what it traded off.
Consensus · Claims & Evidence Table
Consensus
The Claims & Evidence Table, added in August 2025, lists an answer's key claims, each with a color-coded strength bar labeled, for example, Strong or Moderate, and chips for the papers behind it; hovering a citation shows the exact quote it rests on. On yes/no questions the Consensus Meter adds a Yes / Possibly / Mixed / No split of the relevant papers among the top 20, with their count (N), and appears only when at least 5 qualify; each cited paper is colored by its stance. Both ship in Pro answers (10 a month on the free plan), and the table rates key claims when the question suits it or when asked, not every sentence of the prose.
Visit Consensus ↗
Scite · Smart Citations
Scite (Research Solutions)
A deep-learning model labels each citation statement Supporting, Contrasting or Mentioning, and a paper's Report counts each label, plus an Unclassified bucket, in its Citation Types filter. Each statement is quoted in its surrounding text with the section it sits in, the label and the model's confidence in that label (for example "confidence: 89%"). The Flag classification button lets a reader propose a different label; if two expert reviewers agree, it is accepted and marked "Expert classified". The signal grades how later papers cite a published study, not the claims in an AI-written answer.
Visit Scite (Research Solutions) ↗
Research Frame
Why this pattern exists, and where it breaks.
Attach visible confidence, sourcing, and uncertainty cues to each claim.
Plain answers flatten strong and weak claims into the same voice.
Users may over-trust the badge system itself or mistake style for rigor.
Common questions
What people ask about this pattern.
When should I use the Confidence Signals pattern?
+
Answers mix high-evidence and low-evidence claims and the reader cannot tell them apart.
When should I avoid the Confidence Signals pattern?
+
Reliable confidence estimates do not exist and the signal would be decorative — or the reader has no lever to pull, where a confidence band transfers anxiety without transferring any control to act on it.
What problem does Confidence Signals solve?
+
Confident wording hides uneven evidence quality and uncertainty.
Why is chat the wrong fit for this?
+
Plain answers flatten strong and weak claims into the same voice.
Grounded in Research
Not just opinion.
The papers, guidelines, and usability studies this pattern rests on.
Don Norman
Basic Books · revised 2013 edition · 1988
The feedback and visibility principles every interface inherits. Output Undo, Action Audit, and most Observability patterns are direct applications of "the system must reflect its state back to the user."
Google PAIR team
Google Research · 2021 · refreshed 2026-07
Six-chapter first-party guide. Its Explainability + Trust and Feedback + Control chapters map almost one-to-one onto our Trust & Verification and control patterns; its confidence-display guidance ("determine if, when, and how to show confidence") is the softer default our confidence-signals avoidWhen encodes.
Raluca Budiu, Feifei Liu, Emma Cionca, Amy Zhang
Nielsen Norman Group · 2023
Diary study of real chatbot use: only ~4% of turns include a clarifying question and only ~22% of AI answers get fact-checked — the empirical backbone for asking-before-guessing and for making verification cheap enough to actually happen.
Related Patterns
Curated connections, not just neighbors.
Hand-picked relationships — what this pattern pairs with in practice, what it replaces, what it builds on, and what builds on it.
Often paired with
Citation Trail
Confidence Signals say how sure each claim is; Citation Trail shows on hover which source span backs each sentence.
Often paired with
Disambiguation Branch
Disambiguation Branch settles what you meant before the run; Confidence Signals then show how well each claim holds up.