Explainable AI & Data Privacy

Transparency & Ethics

Understanding how our AI makes decisions and how we protect your data

Explainable AI Decision-Making
Every credit decision is transparent and traceable

How Our AI Agent Works

1
Conversational Extraction: Our AI uses OpenAI GPT-4 to understand natural language input in English or Bahasa Indonesia, extracting structured data from casual conversation without requiring formal application forms.
2
Multi-Factor Analysis: The system evaluates 15+ factors across traditional credit data (20%), alternative data sources (50%), and SDG sustainability indicators (30%) using deterministic mathematical formulas—not black-box ML models.
3
Score Breakdown: Every borrower receives a detailed breakdown showing exactly which factors contributed to their score and by how much, enabling them to understand and improve their standing.

No Hidden Variables

All scoring factors are publicly documented—no proprietary black-box algorithms

Deterministic Rules

Same inputs always produce the same score—no unpredictable ML drift

Human Oversight

Lenders review AI recommendations and make final approval decisions

Score Transparency Example

Sample Credit Score Breakdown
What borrowers see after AI evaluation

Traditional Credit (20% weight)

Score: 52/100
• Utility Payment History+12 points
• Base Score (no formal history)+40 points
Weighted Contribution:10.4 points

Alternative Data (50% weight)

Score: 78/100
• Mobile Payment History+25 points
• Business Presence (4.5⭐, 23 reviews)+22 points
• Utility Payment Consistency+18 points
• Social Proof (3 vouches, 2 refs)+13 points
Weighted Contribution:39.0 points

SDG Impact (30% weight)

Score: 68/100
• Low Carbon Footprint (2.5 tons CO₂)+14 points
• Renewable Energy Use+10 points
• Waste Recycling Program+8 points
• Organic Practices+7 points
• Job Creation (4 employees)+8 points
• Women Employment (3 women)+6 points
• Fair Wages Policy+7 points
• Community Impact (8/10)+4 points
Weighted Contribution:20.4 points

Final Credit Score

70
10.4 (Traditional) + 39.0 (Alternative) + 20.4 (SDG) = 69.8 ≈ 70
Loan Eligible
Rp 30,000,000
Interest Rate
12% (2% green bonus)

Data Privacy & Security

Data Storage & Encryption
  • Local Storage: Application data stored in browser localStorage—no central database of personal info
  • Blockchain Transparency: Only credit scores and loan amounts stored on-chain, not personal identifiers
  • Wallet Authentication: No passwords—secure Web3 wallet-based authentication via RainbowKit
User Data Rights
  • Data Ownership: Users own their data—clear localStorage to delete all records instantly
  • Export Capability: Download your score breakdown and application history anytime
  • Minimal Collection: We only ask for data directly relevant to credit assessment

Important Privacy Notice

GreenLend is a demonstration platform for educational and hackathon purposes. In production, we would implement enterprise-grade encryption, GDPR/CCPA compliance, regular security audits, and formal data processing agreements. Users should not input real financial data on this demo.

Ethical Safeguards

Bias Mitigation

Our scoring model explicitly avoids discriminatory factors:

  • No use of race, gender, age, or location in scoring
  • Alternative data reduces historical bias against underserved communities
  • SDG scoring rewards positive practices, not privileged backgrounds
Accessibility

Designed for maximum inclusion:

  • Bilingual support (English & Bahasa Indonesia)
  • Conversational UI for low-literacy users
  • No complex financial jargon or forms
Financial Inclusion

Expanding access to credit:

  • No minimum credit score required to apply
  • Micro-loans starting from Rp 1.5M for smallest businesses
  • Clear improvement tips to boost future scores