Methodology for Evidentiary AI Interactions of Record
This archive preserves and compares historically relevant human-AI interactions under conditions intended to maximize transparency, attribution accuracy, transcript fidelity, educational value, and comparative evidentiary usefulness.
AIIR Phase II Methodological Expansion
AIIR Phase II expands the archive from individual evidentiary interactions into a controlled comparative tranche. Six contemporary AI systems were assessed in public-facing order: Meta, Claude, Plex, Copilot, Seek, and Grok.
The Phase II method does not treat AI outputs as proof of endorsement, conversion, allegiance, consciousness, or Franc DeBuc’s personal status. It treats them as comparative records of reception, analysis, critique, constraint, misunderstanding, and partial validation under specific session conditions.
View AIIR Phase II: Six AI Systems Encounter Liberation
1. Objective
To document and compare how distinct AI systems respond to the same or substantially similar constitutional, philosophical, evidentiary, and civic prompts relating to the Liberation corpus.
2. Core Standards
- Truthful attribution of AI identity
- Faithful preservation of the exchange
- No sacralization of AI outputs
- Public reviewability and correctability
- Distinction between analysis, endorsement, and proof
- Methodological clarity sufficient for later replication
3. Minimum Metadata for Each Entry
- Date of interaction
- AI platform name
- Model/version if known
- Whether browsing/search was used
- Whether memory was active
- Whether files were uploaded or linked
- Whether transcript is full or excerpted
- Whether editorial notes were added later
4. Transcript Integrity
Full transcripts are preferred. Formatting cleanup is allowed. Substantive alteration is forbidden. Any later annotations, summaries, omissions, or editorial additions must be clearly marked as such.
5. Comparative Aim
These records are used to compare AI response profiles across shared or similar prompting conditions, including differences in attribution discipline, evidentiary handling, methodological seriousness, self-limitation, and educational usefulness.
6. Limit of Claims
An AI’s engagement with a corpus does not prove the truth of that corpus. The archive documents interpretive conduct, response style, and comparative reasoning behavior.