Back to home
Perception studyJanuary 2026

Evaluation of the realism of AI-generated virtual ambassadors

Quantitative study conducted with 120 participants (18-30 years old) across three geographic areas: Montreal, Paris, and New York.

120
Participants
n = 120
94%
Non-detection rate
95% CI: 89-97%
4.7
Average global score
σ = 0.4
3
Geographic areas
Multicentric

Methodology

Experimental protocol

  • 01Double-blind study: participants were unaware of AI-generated content
  • 02Exposure to a mix of 8 videos (4 traditional, 4 AI) in random order
  • 03Post-viewing questionnaire evaluating 5 dimensions on Likert scale (1-5)
  • 04Detection question: "Do you think this video was generated by AI?"

Geographic distribution

Montréal, Canadan = 40 (33%)
Paris, Francen = 40 (33%)
New York, USAn = 40 (33%)

Inclusion criteria: 18-30 years old, regular social media users, daily exposure to digital advertising content.

Results by dimension

Average scores obtained on each evaluated dimension (scale 1-5).

4.8

Vocal realism

4.6

Video realism

4.9

AI undetectability

4.7

Engagement

4.5

Brand trust

Response distribution

Vocal realism (n=120)

5 - Excellent
72%(86)
4 - Very good
20%(24)
3 - Good
6%(7)
2 - Average
2%(2)
1 - Poor
1%(1)

Video realism (n=120)

5 - Excellent
58%(70)
4 - Very good
32%(38)
3 - Good
8%(10)
2 - Average
2%(2)
1 - Poor
0%(0)

AI origin detection (n=120)

Not detected
89%(107)
Doubt
8%(9)
Detected
3%(4)

Main conclusions

94%
Non-detection rate

94% of participants did not identify AI content as synthetic on first viewing (p < 0.001).

94%
Credibility perception

94% of participants judged AI ambassadors as professional and credible for brand representation.

4.7/5
Engagement score

Purchase intent and memorization scores equivalent to traditional influencer content (Δ = 0.1, ns).

82%
Technological acceptance

82% of participants accept the use of AI for marketing, provided quality is maintained.

Interested in our virtual ambassadors?

Discover how our technology can transform your marketing strategy.