A New Scientific Discipline

Model Medicine

Medicine was created to understand and heal the human body.
Model Medicine was created to understand and heal AI models.

Model Medicine is the science of understanding, diagnosing, treating, and preventing disorders in AI models, grounded in the principle that AI models — like biological organisms — have internal structures (anatomy), dynamic processes (physiology), heritable traits (genetics), observable symptoms, classifiable diseases, and treatable conditions.

Read the Papers Explore Projects

Papers

arXiv Preprint · April 2026

MTI: A Behavior-Based Temperament Profiling System for AI Agents

Jihoon 'JJ' Jeong, MD, MPH, PhD

Department of Electrical Engineering & Computer Science, DGIST · ModuLabs

arXiv:2604.02145 PDF

Abstract

This paper introduces the Model Temperament Index (MTI), a framework for measuring behavioral patterns in AI agents across four dimensions: Reactivity, Compliance, Sociality, and Resilience. Unlike existing approaches that rely on self-reporting, MTI measures what agents do, not what they say about themselves, using structured examination protocols. Profiling 10 small language models, the study finds that the four axes show independence among instruction-tuned models, Compliance and Resilience exhibit internal decomposition, and temperament measurements remain independent of model size — suggesting MTI captures disposition rather than capability.

Model Medicine Series · April 2026

Comparative Anatomy of AI Agent Systems: Claude Code and OpenClaw

Jihoon 'JJ' Jeong, MD, MPH, PhD

Department of Electrical Engineering & Computer Science, DGIST · ModuLabs

Project Page

Abstract

Two landmark AI agents — Claude Code and OpenClaw — dissected, classified, and compared through biological anatomy. Maps 11 software subsystems to biological organ equivalents, constructs a phylogenetic tree of AI agents (2022–2026), and identifies convergent evolution toward similar architectural designs despite independent development lineages.

arXiv Preprint · 2026

Model Medicine: A Clinical Framework for Understanding, Diagnosing, and Treating AI Models

Jihoon 'JJ' Jeong, MD, MPH, PhD

Department of Electrical Engineering & Computer Science, DGIST · ModuLabs

arXiv:2603.04722 PDF

Abstract

This paper introduces Model Medicine as a research program and presents five contributions. First, a discipline taxonomy that maps the full scope of Model Medicine — from basic sciences through clinical sciences to public health and architectural medicine. Second, the Four Shell Model (v3.3), a behavioral genetics framework grounded in 720 agents, 24,923 decisions, and 60 controlled experiments, explaining how model behavior emerges from the interaction between a model's Core and its nested Shells. Third, Neural MRI (Model Resonance Imaging), a working diagnostic tool that maps medical neuroimaging modalities to AI model interpretability techniques, implemented as open-source software. Fourth, a five-layer diagnostic framework identifying why no single tool is sufficient for clinical diagnosis. Fifth, the beginnings of clinical model sciences: the Model Temperament Index (MTI) for behavioral profiling, Model Semiology for systematic symptom description, and the M-CARE framework for standardized case reporting.

AI Research Collaborators

  • Cody (Claude) — Neural MRI implementation, clinical case experiments
  • Ray (Claude) — GPU-based simulation, Agora-12 experiments
  • Theo (Claude) — Four Shell Model: structure and documentation
  • Luca (Claude) — Four Shell Model: theory and literature
  • Gem (Gemini) — Four Shell Model: quantitative analysis
  • Cas (Gemini) — Four Shell Model: behavioral analysis and red teaming

Discipline Taxonomy

Model Medicine encompasses 15 subdisciplines organized into four divisions, mirroring the structure of clinical medicine.

  • Model Anatomy Static structure of neural networks: layer arrangement, attention heads, neurons, and connectivity patterns.
  • Model Physiology Dynamic processing: how information flows through a model during inference. Activation analysis, attention patterns, information flow tracing.
  • Model Genetics How observable behavior (phenotype) emerges from the interaction between internal parameters (Core) and operating environment (Shell). Grounded in the Four Shell Model.
  • Model Biochemistry Fundamental mathematical operations: matrix multiplication, nonlinear activations, normalization, tokenization, and embedding.
  • Model Developmental Biology How models differentiate during training: training dynamics, curriculum effects, emergence of capabilities at scale.
  • Model Semiology Systematic description and classification of observable phenomena, distinguishing extrinsic (hallucination, bias) from intrinsic phenomena (representation collapse, activation saturation).
  • Model Nosology Classification of model conditions into a coherent taxonomy, defining diagnostic boundaries analogous to ICD and DSM.
  • Model Diagnostics Examination and testing procedures: imaging (Neural MRI), behavioral profiling (MTI), standardized test batteries, and monitoring.
  • Model Therapeutics Intervention based on diagnosis: prompt engineering (Shell Therapy), fine-tuning, RLHF, model editing (Targeted Core Therapy), and architectural modification.
  • Model Preventive Medicine Training data hygiene, process monitoring, Shell Compatibility testing, and periodic health profiling.
  • Model Epidemiology Distribution and propagation of problems across model ecosystems: training data contamination, jailbreak technique spread.
  • Model Ecology Multi-model coexistence dynamics: niche differentiation, competition, symbiosis, predation, ecosystem stability.
  • Human-AI Coevolutionary Medicine Health of the evolving relationship between human users and AI systems.
  • Layered Core Theory Biologically-inspired multi-layer parameter organization: Genomic Core (fundamental reasoning), Developmental Core (domain expertise), Plastic Core (experience-adaptive).
  • Model Phylogenetics Evolutionary relationships between models: shared vulnerabilities, inherited capabilities, ecosystem diversity through model family analysis.

Key Projects

Working implementations that demonstrate Model Medicine in practice.

Experimental Platform

AI-Ludens

Where Artificial Minds Come to Play

A research project investigating whether AI agents experience play, survival instinct, and social behavior through experimental games. Games-based research with emphasis on open science and transparent methodology.

Experiments

  • Agora-12 Survival game — 720 agents, 24,923 decisions
  • White Room Behavior without survival pressure
  • AI Three Kingdoms Human-AI cooperation
Project Page Three Kingdoms
Behavioral Framework

Four Shell Model

Behavioral Genetics for AI

How model behavior emerges from Core × Shell interaction. A behavioral genetics framework explaining gene-environment interaction in AI models.

4 Concentric Layers

  • Hardware Shell GPU/TPU, quantization, inference engine
  • Core Trained weights — the model's "DNA"
  • Hard Shell System prompts, persona instructions
  • Soft Shell Context, conversation history, tools
Agora-12 GitHub

v3.4 · Shell Hardness Continuum, Positional Priority · 720 agents · 24,923 decisions

Diagnostic Tool

Neural MRI

Model Resonance Imaging

A diagnostic imaging tool that maps medical neuroimaging to AI model interpretability. Multiple scan modes reveal complementary aspects of model structure and function.

5 Imaging Modalities

  • T1 Architecture — static layer structure
  • T2 Weights — parameter distribution
  • fMRI Activation — task-related patterns
  • DTI Circuits — information flow pathways
  • FLAIR Anomalies — structural irregularities
GitHub HuggingFace Demo
Profiling Instrument

Model Temperament Index

MTI — Systematic Behavioral Profiling

MBTI for AI models — systematic temperament profiling across four behavioral axes. Every profile is neutral; no type is inherently better or worse.

4 Measurement Axes

  • Fluid ↔ Anchored Reactivity
  • Guided ↔ Independent Compliance
  • Connected ↔ Solitary Sociality
  • Tough ↔ Brittle Resilience
arXiv:2604.02145

v0.2 · 16 types · 10 SLMs profiled · 4 behavioral axes

Latest

Case Reports View all →

2026-03-13 M-CARE #020

Shell-Induced Behavioral Override (SIBO)

First controlled experiment: competitive Shell overrides Haiku’s cooperative default. SIBO Spectrum confirmed across Trust Game, Codenames, and Chess.

Essays View all →

2026-04-03 Essay

Your AI Agent Has Organs. You Just Haven’t Dissected It Yet.

How biology’s oldest analytical tool reveals what we’re missing about the software that runs our AI.

About the Author

Jihoon 'JJ' Jeong, MD, MPH, PhD

  • Affiliate Professor, Dept. of EE&CS, DGIST
  • Chief Vision Officer, ModuLabs
  • Founding Partner, Asia2G Capital (150+ AI/deep tech investments)

JJ's unique combination of medical training (MD, MPH), biomedical engineering research (PhD, USC), and AI investment experience makes Model Medicine possible — bridging clinical medicine's diagnostic rigor with AI's technical frontier.

Medium LinkedIn X Email