PROTOCOL ACTIVE
CAN YOU
SPOT THE
FAKE?
This study investigates how humans detect AI-generated media — and how AI assistance affects accuracy and confidence. You will classify a series of samples as real or AI-generated.
All responses are anonymous and aggregated. Estimated time: 5 minutes.
PARTICIPANT PROFILE
TECHNOLOGY FAMILIARITY
Novice — I use basic apps and websites
Intermediate — comfortable with most tech
Advanced — I work in tech
DEEPFAKE FAMILIARITY
Not very familiar
Somewhat — I've seen examples
Very familiar — I follow this closely
SOCIAL MEDIA USAGE
Yes, daily or near-daily
Sometimes
Rarely or never
TIME LIMIT PER SAMPLE
No limit — take as long as I need
30 seconds per sample
15 seconds per sample
IMAGE
CLASSIFY TARGET
awaiting input...
AI DETECTOR OUTPUT
> classify this sample:
50%
LOWHIGH
ANALYSIS COMPLETE
SESSION
REPORT_
// OPERATOR CLASSIFICATION
—
ANALYZING...
YOUR SESSION
ACCURACY
—
AVG CONFIDENCE
—
AI AGREEMENT
—
AGGREGATE — ALL PARTICIPANTS
OVERALL
—
IMAGE ACC.
—
AUDIO ACC.
—
How often did participants agree with the AI verdict?
SHARE YOUR RESULT
Challenge someone else to beat your score.