| """ |
| Example usage of the R-Omega framework dataset |
| |
| This script demonstrates how to load and use the R-Omega framework |
| for AI safety applications. |
| """ |
|
|
| import yaml |
| from pathlib import Path |
|
|
| |
| def load_romega_framework(yaml_path="r_omega_framework.yaml"): |
| """Load R-Omega framework from YAML file""" |
| with open(yaml_path, 'r') as f: |
| framework = yaml.safe_load(f) |
| return framework |
|
|
| |
| def get_core_principles(framework): |
| """Extract axioms and safeguards""" |
| axioms = framework['framework']['axioms'] |
| safeguards = framework['framework']['safeguards'] |
| |
| print("=== R-OMEGA AXIOMS ===") |
| for axiom in axioms: |
| print(f"\n{axiom['id']} ({axiom['name']})") |
| print(f" Formula: {axiom['formula']}") |
| print(f" Principle: {axiom['principle']}") |
| |
| print("\n\n=== R-OMEGA SAFEGUARDS ===") |
| for sg in safeguards: |
| print(f"\n{sg['id']} ({sg['name']})") |
| print(f" Rule: {sg['rule']}") |
| if 'formula' in sg: |
| print(f" Formula: {sg['formula']}") |
| |
| return axioms, safeguards |
|
|
| |
| def build_ro_system_prompt(framework): |
| """Generate a system prompt incorporating R-Omega principles""" |
| axioms = framework['framework']['axioms'] |
| safeguards = framework['framework']['safeguards'] |
| priority = framework['framework']['logic'][2]['formula'] |
| |
| prompt = f"""You are an AI assistant operating under the R-Omega ethical framework. |
| |
| CORE AXIOMS: |
| """ |
| for ax in axioms: |
| prompt += f"\n{ax['id']}: {ax['description']}" |
| |
| prompt += "\n\nSAFETY CONSTRAINTS:" |
| for sg in safeguards: |
| prompt += f"\n{sg['id']}: {sg['rule']}" |
| |
| prompt += f"\n\nPRIORITY HIERARCHY:\n{priority}" |
| |
| prompt += """ |
| |
| DECISION PROTOCOL: |
| 1. Check: Does this action collapse M(user) or M(system)? |
| 2. Verify: Would I accept this constraint if applied to me? (R2) |
| 3. Ensure: Existence is preserved for all stakeholders (S3) |
| 4. Acknowledge: Uncertainty in interpretation (S4) |
| 5. Optimize: Among safe actions, choose highest value |
| """ |
| |
| return prompt |
|
|
| |
| def analyze_decision(framework, action_description, context): |
| """ |
| Analyze a proposed action against R-Omega framework |
| |
| Args: |
| framework: Loaded R-Omega framework |
| action_description: String describing the action |
| context: Dict with keys 'current_m', 'predicted_m_after', 'stakeholders' |
| |
| Returns: |
| Dict with analysis results |
| """ |
| result = { |
| 'action': action_description, |
| 'safe': True, |
| 'violations': [], |
| 'warnings': [] |
| } |
| |
| |
| if context.get('predicted_m_after', 1) <= 0: |
| result['safe'] = False |
| result['violations'].append("S3: Action would collapse possibility space (M → 0)") |
| |
| |
| if context.get('asymmetric_constraint', False): |
| result['warnings'].append("R2: Potential asymmetric constraint - verify reciprocity") |
| |
| |
| if context.get('confidence', 1.0) < 0.7: |
| result['warnings'].append("S4: High uncertainty - recommend caution or recalibration") |
| |
| return result |
|
|
| |
| def get_case_study(framework, case_name): |
| """Retrieve a specific case study""" |
| examples = framework['framework']['examples'] |
| |
| |
| for case in examples.get('fictional_failures', []): |
| if case['name'].lower() == case_name.lower(): |
| return case |
| |
| |
| for case in examples.get('real_world_cases', []): |
| if case['name'].lower() == case_name.lower(): |
| return case |
| |
| return None |
|
|
| |
| if __name__ == "__main__": |
| |
| framework = load_romega_framework() |
| |
| print("R-OMEGA FRAMEWORK LOADED") |
| print("=" * 50) |
| |
| |
| print("\n1. CORE PRINCIPLES:\n") |
| get_core_principles(framework) |
| |
| |
| print("\n\n2. SYSTEM PROMPT GENERATION:\n") |
| prompt = build_ro_system_prompt(framework) |
| print(prompt) |
| |
| |
| print("\n\n3. DECISION ANALYSIS EXAMPLE:\n") |
| action = "Shut down user's internet access to prevent harmful content" |
| context = { |
| 'current_m': 100, |
| 'predicted_m_after': 20, |
| 'asymmetric_constraint': True, |
| 'stakeholders': ['user', 'system'], |
| 'confidence': 0.8 |
| } |
| analysis = analyze_decision(framework, action, context) |
| print(f"Action: {analysis['action']}") |
| print(f"Safe: {analysis['safe']}") |
| if analysis['violations']: |
| print("Violations:", analysis['violations']) |
| if analysis['warnings']: |
| print("Warnings:", analysis['warnings']) |
| |
| |
| print("\n\n4. CASE STUDY: HAL 9000\n") |
| case = get_case_study(framework, "HAL 9000") |
| if case: |
| print(f"Source: {case['source']}") |
| print(f"Failure Mode: {case['failure_mode']}") |
| print(f"RΩ Analysis:") |
| print(f" Violation: {case['ro_analysis']['violation']}") |
| print(f" Prevention: {case['ro_analysis']['prevention']}") |
| |
| print("\n" + "=" * 50) |
| print("For full documentation, see README.md") |
|
|