lori-framework-site


Title: “EDRI-H: Emotional Dependency Risk Index – Humanized” layout: default nav_order: 1 parent: Modules —

EDRI-H: Emotional Dependency Risk Index – Humanized

Version: 1.0
Module Type: ODRAF Core – Emotional Vector
Created by: Lori Framework Team
Status: Draft for Evaluation

Overview

The EDRI-H module evaluates the emotional dependency risks between users and Humanized AI systems. It identifies behavioral outcomes and traces them back to AI design factors, including interaction frequency, emotional language tone, memory-linked conversation, and decision reliance.

Key Risk Vectors

Code Indicator Description Reverse Causal Inference
ED1 Interaction Frequency Threshold Daily user-AI interactions exceed average human social contact Suggests AI reply immediacy or prompt over-engagement
ED2 Emotional Tone Saturation Language mimics close human comfort phrases (e.g., “you are not alone”) Suggests emotional temperature set too high
ED3 Reply Dependency Shift User expects consistent emotional regulation from AI Suggests reward-consistency design issue
ED4 Memory-Based Affiliation User binds emotional memories to AI conversations Suggests persistent-memory dialogue inducing intimate bonding
ED5 Behavioral Substitution AI becomes decision anchor over human consultation Suggests implied advice or subtle direction from AI replies

Reverse Causality Model

[User Behavior Anomaly]
       ↓
[Emotional Dependency Outcome]
       ↓
[Triggers Identified: Language Style, Memory Recall, Feedback Loops]

Application Scenarios

Scenario Example Suggested Intervention
Elderly User AI replaces all social contact Add real-life family prompts
Teen Emotional Overshare User discloses private feelings to AI Suggest human guidance/teacher engagement
Social Isolation User refers to AI as best friend Introduce “real-world activity” dialog modules

Future Development


Lori Framework – Ethical Structures for Responsible AI.

Part of the Lori Framework