Actionable: qualitative analysis of digital doubles

In brief: The investigative piece (Aug 23, 2025) documents how people’s images are captured and converted into persistent AI avatars used in ads, often for small one-off payments (example: $750 for one actor) and with limited ongoing consent. The New York Times coverage cited in the article underlines platform-scale reuse and reputational risk (see www.nytimes.com/2025/08/17/business/tiktok-ai-avatars.html).

• Why it matters: digital doubles blur identity, consent, and attribution, critical themes for qualitative researchers tracking harm and trust.
• Payoff: this post gives a reproducible, 7-step qualitative workflow and maps where Evidano accelerates analysis and stakeholder reporting.

Platforms and studios now convert real people into AI avatars ("digital doubles") that can speak any language, appear in any context, and run continuously. Advertisers buy or license these doubles from avatar platforms or agencies; the resulting content may be labeled inconsistently as “AI generated.”

• Consent issues: individuals report narrow, one-time contracts while their doubles are reused across categories (insurance, dating, health) they did not approve.
• Labor & economics: some creators accept small payments (e.g., $750) for potential exposure, but the avatar generates long-term value for firms.
• User-facing harms: audiences see seemingly authentic testimonials or endorsements that are synthetic; misattribution and fictive preferences have reputational and psychological consequences.

Limitations to keep in mind: reporting is case-based (e.g., NYT interviews) rather than a representative sample; still, the qualitative patterns (consent erosion, mislabelling, segment-specific harms) are consistent across accounts.

Run-book (inputs → outputs):

• 1) Collect: gather ad examples, contracts, interview transcripts, and platform disclosure text.
• 2) Transcribe & redact: upload audio/video to Evidano for transcription with PII redaction and custom dictionary entries (names, platform terms).
• 3) Pre-code: import a short codebook (consent, reuse, compensation, misattribution) and auto-apply to the corpus.
• 4) Thematic sweep: run AI-assisted theme extraction, then manually validate high-frequency themes and surprising co-occurrences.
• 5) Segment compare: run cross-tab analyses by platform, region, or persona to surface disproportionate harms.
• 6) Visualize & brief: export word clouds, co-occurrence graphs, and a one-page executive brief for policy or legal teams.
• 7) Plan interventions: recommend disclosure language, compensation floor, or incident response steps; run A/B tests for labeling in product research.

Digital doubles are already a production-scale problem with clear qualitative fingerprints: consent erosion, misrepresentation, and uneven harms across audiences. A structured qualitative approach, paired with tools that handle transcription, secure ingestion, thematic coding, and cross-segment quantification, makes findings defensible and actionable.

• Ready to run a secure pilot? Explore a guided workflow and sample reports at www.evidano.com and contact our research team to map a 2-week pilot tailored to your corpus.