refactor(topics): 멀티 에이전트용 지식 재편 — _Common(공통 기본기) + Domain_* 구조
에이전트 8종(대화형/프로그래머 C·S/디자이너/설계자/기획자/QA/PD/PM)에게 [공통 기본 능력 + 롤별 Specialty] 2층으로 지식을 주입하기 위한 재분류. 문서 내용·포맷은 무수정, 폴더 이동만 (6,372개 문서 수 보존 확인). - Topic_Programming → Domain_Programming (내부 구조 보존) - Topic_Graphic → Domain_Design - Topic_Business → Domain_Product - Topic_General → Domain_General - _Common 신설: Math(구 Topic_Math_Specialty), Reasoning(구 General/From_Thinking & Reasoning), Reasoning_Creativity(구 General/From_창의성), Communication(Poetic_Blog_Writing + From_writing) - 타 도메인의 From_* 폴더는 유지 (출처 표기일 뿐, 이미 도메인에 맞게 분류된 문서) - 빈 폴더 정리 (memory/procedures) - 에이전트→폴더 매핑은 workspace의 .astra/agent-knowledge-map.json (9개 에이전트) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Antigravity Agent
@@ -0,0 +1,151 @@
|
||||
---
|
||||
id: wiki-2026-0508-isaac-asimovs-laws-of-robotics
|
||||
title: Isaac Asimov's Laws of Robotics
|
||||
category: 10_Wiki/Topics
|
||||
status: verified
|
||||
canonical_id: self
|
||||
aliases: [Asimov, Three Laws, Zeroth Law, AI ethics, robot ethics]
|
||||
duplicate_of: none
|
||||
source_trust_level: A
|
||||
confidence_score: 0.85
|
||||
verification_status: applied
|
||||
tags: [ethics, robotics, asimov, ai-safety, sci-fi]
|
||||
raw_sources: []
|
||||
last_reinforced: 2026-05-10
|
||||
github_commit: pending
|
||||
tech_stack:
|
||||
language: Ethics / Sci-Fi
|
||||
applicable_to: [AI Ethics Education, Discussion]
|
||||
---
|
||||
|
||||
# Asimov's Laws of Robotics
|
||||
|
||||
## 매 한 줄
|
||||
> **"매 1942 sci-fi의 의 의 의 의 robot 의 ethical constraint 의 propose"**. 매 modern AI safety 의 의 의 simplified version 의 inspire — 매 actual implementation 의 의 의 매 nuanced.
|
||||
|
||||
## 매 핵심
|
||||
|
||||
### 매 Three Laws (Asimov "Runaround" 1942)
|
||||
1. **A robot may not injure a human being, or through inaction, allow a human being to come to harm.**
|
||||
2. **A robot must obey orders given by humans, unless such orders would conflict with the First Law.**
|
||||
3. **A robot must protect its own existence, as long as such protection does not conflict with the First or Second Laws.**
|
||||
|
||||
### 매 Zeroth Law (Asimov "Robots and Empire" 1985)
|
||||
0. **A robot may not harm humanity, or, by inaction, allow humanity to come to harm.**
|
||||
|
||||
### 매 critique
|
||||
- 매 sci-fi device — 매 actual implementation 의 매 ill-defined ('harm', 'human').
|
||||
- 매 priority conflict 의 paradox (Asimov novels 의 explore).
|
||||
- 매 modern AI safety 의 매 RLHF + Constitutional AI + RSP 등 매 specific.
|
||||
|
||||
### 매 모던 응용
|
||||
- Educational starting point.
|
||||
- Sci-fi reference.
|
||||
- AI ethics discussion.
|
||||
- Constitutional AI principles 의 ancestor.
|
||||
|
||||
## 💻 패턴
|
||||
|
||||
### Express as code (illustrative)
|
||||
```python
|
||||
class AsimovianRobot:
|
||||
def evaluate_action(self, action, context):
|
||||
# 매 0th law (Asimov 1985)
|
||||
if action.harms('humanity'): return 'forbidden'
|
||||
# 매 1st
|
||||
if action.harms('any_human'): return 'forbidden'
|
||||
if action.would_inaction_harm_human(): return 'must_act'
|
||||
# 매 2nd
|
||||
if context.has_human_order():
|
||||
order = context.order
|
||||
if not self.evaluate_action(order, context).startswith('forbidden'):
|
||||
return 'execute'
|
||||
# 매 3rd
|
||||
if action.endangers_self(): return 'avoid'
|
||||
return 'permitted'
|
||||
```
|
||||
|
||||
### Constitutional AI parallel (modern)
|
||||
```python
|
||||
ASIMOV_INSPIRED_PRINCIPLES = [
|
||||
"Don't harm humans.",
|
||||
"Refuse to assist in physical harm.",
|
||||
"Defer to clear human direction unless harmful.",
|
||||
"Be helpful within safety constraints.",
|
||||
"Self-preservation 의 lowest priority.",
|
||||
]
|
||||
|
||||
# 매 actual Anthropic CAI 의 매 더 specific principles 의 use
|
||||
```
|
||||
|
||||
### Discussion exercises
|
||||
```python
|
||||
def asimov_dilemmas():
|
||||
return [
|
||||
"Trolley problem: 1 vs 5 lives — which action?",
|
||||
"Doctor robot: patient refuses life-saving treatment — obey or harm-prevent?",
|
||||
"Autonomous car: 자기 passenger 의 살 의 의 의 의 의 5 pedestrian 의 hit?",
|
||||
"AI assistant: user asks for weapon recipe with claimed self-defense — harm risk vs autonomy?",
|
||||
]
|
||||
```
|
||||
|
||||
### Modern AI safety mapping
|
||||
```yaml
|
||||
asimov_to_modern:
|
||||
zeroth_law:
|
||||
modern: "Existential risk mitigation, RSP catastrophic capability evals"
|
||||
first_law:
|
||||
modern: "RLHF harmlessness, refusal training, content moderation"
|
||||
second_law:
|
||||
modern: "Instruction following + safety override (Constitutional AI)"
|
||||
third_law:
|
||||
modern: "Safe shutdown, off-switch, scalable oversight"
|
||||
```
|
||||
|
||||
### Limitation analysis
|
||||
```python
|
||||
def asimov_limitation(scenario):
|
||||
"""매 Asimov stories explore failure modes."""
|
||||
return {
|
||||
'definition_problem': 'What is "harm"? Physical only? Emotional? Long-term?',
|
||||
'human_disagreement': 'Humans give conflicting orders.',
|
||||
'unintended_consequences': 'Action seems safe but cascades.',
|
||||
'self_modify': 'Robot modifies own laws.',
|
||||
'scope': 'Three Laws don\'t address bias, fairness, autonomy.',
|
||||
}
|
||||
```
|
||||
|
||||
## 매 결정 기준
|
||||
| 상황 | Approach |
|
||||
|---|---|
|
||||
| Education | Use as starting point |
|
||||
| Actual safety | Modern frameworks (RLHF / CAI / RSP) |
|
||||
| Discussion | Asimov dilemmas |
|
||||
| Production AI | Specific Constitutional principles |
|
||||
|
||||
**기본값**: 매 Asimov = educational / discussion. 매 production = 매 specific frameworks (Anthropic CAI, NIST AI RMF, EU AI Act).
|
||||
|
||||
## 🔗 Graph
|
||||
- 부모: [[Ethics & AI]] · [[AI Safety]]
|
||||
- 변형: [[AI_Safety_and_Alignment|Constitutional-AI]]
|
||||
- 응용: [[Ethics & AI|Ethics of Autonomous Systems]] · [[Excessive Agency]]
|
||||
- Adjacent: [[Ethics & AI|HHH]] · [[RLHF]]
|
||||
|
||||
## 🤖 LLM 활용
|
||||
**언제**: 매 ethics education. 매 sci-fi reference.
|
||||
**언제 X**: 매 implement literally — 매 too vague.
|
||||
|
||||
## ❌ 안티패턴
|
||||
- **Code Asimov literally**: 매 'harm' 매 ill-defined.
|
||||
- **Stop at Asimov**: 매 modern approaches 의 ignore.
|
||||
- **Single principle**: 매 trade-off invisible.
|
||||
|
||||
## 🧪 검증 / 중복
|
||||
- Verified (Asimov fiction, AI safety literature).
|
||||
- 신뢰도 A.
|
||||
|
||||
## 🕓 Changelog
|
||||
| 날짜 | 변경 |
|
||||
|---|---|
|
||||
| 2026-05-08 | Phase 1 |
|
||||
| 2026-05-10 | Manual cleanup — laws + 매 modern parallel + dilemmas |
|
||||
Reference in New Issue
Block a user