Show the calculation.
Color values, lightness measurements, subtraction and thresholds make the decision visible.
ONE RULE, TWO VOICES
What makes a system
seem intelligent?
Experience two ways of reading the same colors.
01 / SET THE INPUT
One color profile.
Use a synthetic sample or choose three colors. This experiment classifies color contrast; it does not assess beauty or health.
The sample profiles are synthetic, not photographs or data from participants.
02 / SET THE SESSION
Two short encounters.
Use a code, not a name. Leave blank for an unrecorded walkthrough.
For successive participants, alternate A→B and B→A. Keep the selected colors unchanged across both versions.
Facilitator notes · read before a session
Ask participants to explore each version and answer independently. Do not explain the mechanism or suggest which version should seem more intelligent until the debrief. Use the same neutral wording each time. Participation is optional; do not collect names or photos.
The shared calculation is identical to Assignment 2. Only the result representation changes. The participant questions are exploratory, not validated scales. With a small convenience sample, describe reactions rather than claiming general causal effects.
THE MECHANISM, REVEALED
Different voices.
Exactly the same rule.
Speak as an interpreter.
The numbers are hidden. First-person language turns the same category into a personal-sounding interpretation.
Study question, documentation & credits
How does showing or hiding a simple rule change how people explain, trust and attribute intelligence to its output?
Input colors, L* conversion, thresholds and category are held constant. Transparency and language change together, so this study cannot isolate the effect of either one alone. No artificial delay, seasonal recommendation or extra decision rule is introduced.
Concept: the author’s personal color analysis research. AI collaborator: OpenAI Codex (implementation, parity checks and documentation). Built with HTML, CSS, JavaScript and native browser controls. The sRGB-to-CIELAB L* calculation is reused unchanged from Assignment 2. No external analysis libraries or live AI services. Hosting: OpenAI Sites.
Responses stay in this browser tab’s memory. Download records before refreshing or closing it. No participant reactions have been prefilled.