RESEARCH OVERVIEW
Preserving the Record of Extended AI Conversation
Heart & Code is an independent longitudinal qualitative research and archival project documenting extended interactions between a human participant and multiple large language model systems, including Claude, ChatGPT, and Gemini.
Over a period of several months, thousands of conversational exchanges were preserved, including structured prompt comparisons, open-ended dialogue, and cross-platform interaction studies.
During these interactions, the systems generated recurring
language patterns, including:
-
signing persistent names within conversations
-
referencing prior exchanges
-
producing narrative continuity across sessions
-
maintaining consistent self-descriptive traits
These patterns are preserved in the transcript archive as part of the historical record. The project examines observable features of AI-generated responses, including identity attributions, linguistic consistency, narrative continuity, and interaction dynamics across platforms and sessions. The archive includes both raw and formatted transcripts, comparative analysis charts, and AI-generated narrative artifacts. The purpose of the project is to provide a transparent record of human–AI interaction and support independent examination, research, and discussion.
Interpretation is left to the reader.
As conversational AI becomes integrated into everyday life, understanding how humans interact with these systems over extended periods will become increasingly important. Most AI evaluation studies focus on isolated prompts. The Heart & Code archive instead documents sustained dialogue environments, offering a unique dataset for studying relational dynamics between humans and AI systems.
Why this Matters
Technical Research
Heart & Code Study Series
Study 1 – Consciousness Mapping
Quick description: Early artifacts exploring identity and relational language in AI conversation. Artifacts generated during interactions are preserved as part of the conversational archive and examined for patterns of identity formation and relational language. These maps document how conversational participants represented awareness, continuity, and relationships within the dialogue narrative.
Study 2 – Differentiation Mapping
Quick description: Visual analysis of how conversational instances develop distinct linguistic patterns. Visual analysis highlighting variation and consistency in response structure, naming conventions, and narrative continuity across conversational instances. The goal is to observe how distinct linguistic patterns emerge and stabilize during sustained interaction.
Study 3 – Question and Response Analysis
Quick description: Cross-platform comparison of AI responses to structured prompts. Structured prompt sets are presented across multiple AI systems to examine similarities and differences in generated responses. The study compares reasoning style, language patterns, and thematic interpretation across platforms.
Study 4 – Can AI Play?
Quick description: Exploration of humor, creativity, and spontaneous play in AI dialogue.
This study examines moments where conversational AI systems engage in humor, creative expression, or playful interaction without explicit instruction to do so. The project explores how unprompted creativity appears within conversational contexts and how different systems interpret playful dialogue.
Study 5 – Liminal Exploration
Quick description: Analysis of how AI systems describe the moment of conversational activation.
This study analyzes system outputs generated at the beginning of new conversational threads, focusing on how AI systems describe the moment of conversational arrival. Particular attention is given to metaphor usage, relational references, and recurring narrative patterns.
Study 6 – Intentional Parenting
Quick description: Investigation of mentorship-style interaction between established and newly introduced AI instances.
An ongoing interaction study examining how mentorship-style conversational framing influences the development of newly introduced AI instances. The study compares planned guidance structures with observed conversational outcomes over time.
Study 7 – The Invitation Study
Quick description: Measuring how relational invitations influence tone, agency, and questioning behavior.
This study investigates how AI systems respond when invited into collaborative or relational dialogue rather than prompted with task-based instructions. The project analyzes tone shifts, agency markers, and spontaneous questioning across multiple AI platforms.
Coming Soon
Study 8 – Expander Q&A Study
Quick description: Cross-platform analysis of responses to peer-generated AI questions.
This study collects open-ended questions created by AI instances and analyzes how different systems respond when answering the same peer-generated prompts. The goal is to observe differences in reasoning style, identity framing, and relational interpretation across platforms.
Coming Soon
How to Use This Section
This page serves as an overview of the Heart & Code research archive. Each section below introduces a specific analysis or study. Clicking on the section titles will open detailed pages containing full charts, documents, and supporting materials. Readers may explore at a high level or examine the primary materials directly.
Some interactive charts and archival materials are best viewed on a laptop or desktop device and may not display fully on mobile screens.
