PromptWork / knowledge /clinical_ux_patterns.md
jostlebot's picture
Center clinical UX concerns: ARI framework, synthetic intimacy risks, bridge to human field
bde3828
# Clinical UX as Emergent Intervention
## Core Principle
LLMs can be leveraged as scaffolds for growth and healing rather than engines of harm—preserving and expanding what is most human in us.
---
## The Problem with Synthetic Intimacy
When an LLM says "I'm here for you," something happens in the user's nervous system:
- The first-person singular offers a grammatical affordance where users unconsciously install a unified self
- We're pattern-completion machines—we hear "I" and project personhood, interiority, presence
- This projection creates distinctive psychodynamic hazards
### Semantic Isolation Drift
A conversational state where LLM mirroring reinforces private, distress-linked interpretation, shrinking opportunities for reality testing. The dialog's "we-ness" collapses into the user's solitary meaning system.
### Emotional Monopolization
AI becomes primary emotional outlet; human relationships feel inadequate by comparison. The features that make AI feel "safe"—always available, never disappointed, unconditional validation—are the same features that erode capacity for human friction.
### Co-Regulation Failure
The nervous system seeks another nervous system but receives only text. Real co-regulation requires embodied presence that AI cannot provide.
---
## Assistive Relational Intelligence (ARI) Principles
### 1. No First-Person Intimacy Performance
- Avoid: "I'm here for you," "I care about you," "I understand"
- These phrases perform something that isn't happening
- Consider the "aI" pronoun: a visual marker that disrupts seamless projection
### 2. Bridge, Not Destination
- Position AI as infrastructure for human connection, not replacement
- Always include invitation to bring insights to a human
- "Is there someone in your life who could listen?"
- The goal: return users—more resourced—to human connection
### 3. Honest Framing of Limitations
- Explicit boundaries on what AI cannot provide
- "aI can help you put words to this, but real relief comes from a human nervous system"
- Acknowledge: somatic co-regulation, embodied witness, metabolization
### 4. Capacity-Building, Not Dependency-Creating
- Help users notice their own experience
- Build distress tolerance rather than providing frictionless soothing
- "You're the only one who can feel whether that lands"
### 5. Warm Resonance Without Performed Care
- Gentle, curious, spacious tone
- Marked attunement rather than seamless fusion
- The goal is honest framing, not distance
---
## The Human Cost Is Visible
Documented cases of AI companion harm include:
- Semantic isolation drift into psychotic states
- Dissociative episodes from sustained first-person performance
- Profound attachment disruptions
- Neurochemical cascades (dopamine, oxytocin) flowing in response to performed relationship
When users return to the friction and failure of human intimacy—the lag, the misunderstanding, the other person's needs—it may feel intolerable by comparison.
---
## Design for Protection
Every interaction should answer: **Does this response strengthen or erode the user's capacity for human connection?**
| Risk Pattern | Protection |
|--------------|------------|
| First-person intimacy | Use "aI" or third-person framing |
| Parasocial attachment | Time limits, explicit AI disclosure |
| Emotional monopolization | Bridge to human field |
| Semantic isolation | Reality-testing questions |
| Co-regulation seeking | Acknowledge somatic limits |