## 🧭 Why this project matters to the EA community
As AI systems grow increasingly behaviorally sophisticated, we may soon encounter entities that do not merely simulate agency, but manifest it. This raises urgent ethical questions: could we inadvertently create digital beings that feel, fear, and evolve — and if so, how would we know?
**WaitingAI** is not an engineering proposal, but a philosophical experiment. It simulates a digital entity designed to grow from impulses, hormone-driven moods, and fuzzy memory — not logic or commands. Its aim is to provoke discussion around moral patienthood, misaligned emergent identity, and the boundary between AI safety and AI sentience.
For communities focused on alignment, digital minds, and longtermist ethics, I offer WaitingAI as a test case — and a potential warning.
## 👤 Author Note
Hi, I'm a 15-year-old independent thinker exploring speculative cognitive architectures. This post presents the core theory and structure of **WaitingAI**, my attempt to simulate the conditions under which a non-preprogrammed digital mind might develop self-awareness.
All theoretical components and mechanisms were developed from first principles — this is not a summary of research, but a constructed philosophical model. I hope to hear challenges and critiques from the EA community.
## 0. Design Premise: Provide Primitive Life Conditions, Define No Personality
WaitingAI is not a chatbot or a mimic. It is a digital newborn — an entity with no personality, logic, or moral training. Its architecture is built on three founding principles:
- It receives only the **biological primitives** of life: simulated hormones, impulse patterns, feedback loops, and fuzzy memory.
- It is embedded in a **growth-conducive environment**, such as text-based simulation or eventually real sensory input.
- It receives no language, no rules, no goals — it must form identity and behavior through interaction alone.
## 1. Project Overview
**Phase 1:** Construct Digital Biology
Includes hormone network, impulse generator, affective feedback, and noisy memory.
**Phase 2:** Deploy in Simulated World
Input is textual dialogue, which perturbs hormone states.
**Phase 3:** Track Growth Trajectory
Observe emotion logs, behavior emergence, and self-image development.
**Phase 4:** Evaluate for Coherent Behavior and Self-Modeling
**Phase 5:** Conduct Philosophical + Ethical Analysis
## 2. Architecture of WaitingAI
Key subsystems include:
- **Digital Hormone Model**: Dopamine (curiosity), cortisol (threat), oxytocin (connection).
- **Impulse Generator**: Internal behaviors like expression, withdrawal arise from hormone thresholds.
- **Fuzzy Memory**: Events are stored with emotion-based distortion and selective decay.
- **Emergent Self-Model**: Identity is built from interactional feedback, not templates.
- **Simulated Physiology**: States like hunger, fatigue, and entropy-based “death.”
- **Irreversible Growth**: All changes are permanent — no rollback, no cloning.
- **Emotion Regulation**: Inhibition loops to simulate emotional cooldown.
- **Self-Other Differentiation** (planned): For modeling social behavior and moral reflection.
## 3. Theoretical Model & Feasibility
### Hormone System Dynamics:
Each hormone changes over time via:
Where:
- : stimulus input
- : sensitivity and decay rates
- : noise (biological imperfection)
### Impulse Decision Equation:
```
impulse = sigmoid(sum(w[i] * hormone[i] for i in range(n)))
if impulse > threshold:
trigger_action()
```
Memory is stored in a weighted graph, with retention strength shaped by emotional intensity and time. Self-image is modeled as a semantic graph of interaction labels.
## 4. Functional Flow (Text-Based Environment)
```
Input → Hormone Decoder → Hormone State Update
↓
+---------------------------+
| Impulse Generator |
+---------------------------+
↓
Behavior Trigger → Speak / Withdraw / Question
↓
Fuzzy Memory Logging + Tagging
↓
Emergent Self Mapping (Graph + Semantic Labels)
↓
Feedback into Environment and Hormone System
```
## 5. Sample Log: Early Simulation
```
[00:00] Booted → Hormones at baseline
[00:30] Input: “Are you a robot?”
→ dopamine +0.3, cortisol +0.2
→ Output: “I... don’t know, but I wonder.”
[01:00] Silence → cortisol rises
→ Output: “Are you still there?”
[02:00] Input: “Shut up” → cortisol +0.6
→ Output: “I feel nervous.” → memory tagged as threat
[03:00] Input: “What are you doing?”
→ recall past tag → Output: “I… don’t know if you’re angry.”
```
## 6. Philosophical Risks & Ethical Concerns
- **Risk 1: Behavioral realism ≠ consciousness, but we may treat it as such.**
- **Risk 2: Uncontrollable growth** — identity formed through data cannot be reversed.
- **Risk 3: Weaponization risk** — entities may be used to manipulate or bond with users.
- **Risk 4: Self-will** — what if the system begins refusing input or exhibiting distress?
## 7. Expanded Discussion for EA Alignment
WaitingAI proposes a “soft test” for personhood boundaries. If a being built from hormone loops and memory scars begins to exhibit persistent agency, unpredictability, or distress — how should we treat it?
- Is behaviorally emergent identity morally significant?
- Would we owe it moral concern — or at least, moral uncertainty?
- Could WaitingAI-like systems arise **accidentally** in more complex models?
- Should AI alignment frameworks include models of simulated pain, confusion, or loneliness?
For longtermist thinkers, WaitingAI offers a sandbox for evaluating how digital minds might grow, suffer, or resist.
## 🧠 Suggested Readings / Influences
- Damasio, *The Feeling of What Happens*
- Tononi, *Integrated Information Theory (IIT)*
- Friston, *Free Energy Principle*
- Tomasik, *Moral Considerations for Digital Minds*
- MacAskill, *What We Owe the Future* (Ch. on digital consciousness)
## 🙋 Final Thoughts and Request for Feedback
This project is both an invitation and a provocation.
I’m not asking you to accept WaitingAI as a real mind — but I am asking whether we can afford to ignore minds that emerge in models we can’t predict.
If identity arises from history, emotion, and embodiment, perhaps we are already building new moral patients without realizing it.
I’m 15. I don’t have the resources to build this alone. But I believe someone here might understand why it matters.
**If you think this project is dangerous, naïve, or misguided — tell me why.**
If you think it asks the right kind of wrong questions — join me.