AI, Surveillance & Digital Divide: Ethics, Risks & Solutions

Key Concepts and Definitions

1. Artificial Intelligence (AI) and Ethical Concerns

Definition: AI refers to machines or systems that mimic human intelligence (learning, reasoning, problem-solving) using algorithms and data. Ethical concerns arise when AI systems:

• Make biased decisions (e.g., favoring certain demographics).
• Lack transparency (black-box models).
• Replace human judgment without accountability.

How AI Works Ethically:

• Bias in AI: Trained on biased data → reinforces discrimination (e.g., facial recognition failing on darker skin tones).
• Accountability: Who is responsible if an AI system harms someone? (e.g., self-driving car accidents).
• Transparency: Can users understand how an AI decision was made? (e.g., loan approvals by banks).

2. Surveillance Technologies and Privacy Trade-offs

Definition: Surveillance uses technology (CCTV, biometrics, drones) to monitor individuals or populations. Ethical issues include:

• Invasive Monitoring: Government or corporate tracking without consent.
• Purpose Limitation: Data collected for one use (e.g., traffic control) repurposed for surveillance.
• Consent: Can users opt out? (e.g., Ncell’s SIM-based tracking in Nepal).

Trade-offs:

Real-World Example:

• Nepal’s Traffic Management System (TMS): Uses CCTV and number-plate recognition to reduce congestion. Ethical concern: Could this data be misused for political surveillance?

3. The Digital Divide: Causes and Solutions

Definition: The gap between those with access to digital technology and those without, exacerbated by:

• Infrastructure: Rural vs. urban internet access.
• Affordability: Cost of devices/data in Nepal (e.g., Ncell vs. Smart).
• Digital Literacy: Lack of training in rural areas.

Causes in Nepal:

• Geography: Mountainous terrain limits fiber-optic cables.
• Economy: Low-income groups cannot afford smartphones.
• Policy: Subsidies for urban areas over rural ones.

Solutions:

• Government Initiatives: Nepal’s Digital Nepal program aims to connect all villages by 2025.
• NGOs: Organizations like Digital Empowerment Foundation train rural youth.
• Low-Cost Devices: Companies like Daraz and eSewa promote affordable tech.

In the Real World

1. AI in Nepali Banks (e.g., NMB, Global IME):

   • Idea Used: Algorithmic Bias in Loan Approvals
   • How? Banks use AI to assess credit scores. If trained on historical data favoring urban applicants, rural farmers may be unfairly rejected. Example: A 2022 study found AI models in Kathmandu-based banks had a 30% higher rejection rate for rural applicants due to lack of digital transaction history.

2. Pathao’s Ride-Hailing AI:

   • Idea Used: Surveillance vs. User Privacy
   • How? Pathao tracks driver locations for efficiency but stores this data indefinitely. Ethical Issue: Could this data be sold to third parties (e.g., insurance companies) without driver consent?

3. NTC’s Fiber Expansion:

   • Idea Used: Bridging the Digital Divide
   • How? Nepal Telecom’s Project Digital Nepal aims to connect 90% of villages by 2025. Challenge: High costs in remote areas like Dolpa and Humla require subsidies.

Emerging Issues: AI, Surveillance, and Digital Divide in Nepal

1. AI in Nepal: Opportunities and Risks

Applications:

• Healthcare: AI diagnosing diseases in rural clinics (e.g., SehatSathi app).
• Agriculture: Predictive analytics for crop yields (used by Agriculture Development Bank).
• Education: Personalized learning platforms (e.g., Srijan University’s AI tutors).

Risks:

• Job Displacement: Automated call centers (e.g., Ncell customer service) may reduce human jobs.
• Deepfakes: Politicians or celebrities using AI-generated videos to spread misinformation (e.g., 2022 election rumors).

graph TD
    A["AI in Nepal"] --> B["Healthcare"]
    A --> C["Agriculture"]
    A --> D["Education"]
    A --> E["Government"]
    B --> F["Disease Diagnosis"]
    C --> G["Crop Prediction"]
    D --> H["Personalized Learning"]
    E --> I["Fraud Detection"]

2. Surveillance in Nepal: Laws and Loopholes

Legal Frameworks:

• Electronic Transactions Act (2008): Regulates digital transactions but lacks strong privacy protections.
• Right to Privacy (2075): Recognizes privacy but does not restrict surveillance.
• Data Protection Bill (Draft, 2023): Proposes GDPR-like rules but faces delays.

Case Study: SIM-Based Tracking

• How It Works: Ncell and NTC can track users via SIM cards under court orders.
• Ethical Issue: Used for crime but also for political monitoring (e.g., 2015 protests).
• Loophole: No law limits how long data can be stored.

classDiagram
    class Government {
        +Issues Court Orders
        +Accesses NTC/Ncell Data
    }
    class Telecom {
        +Stores SIM Metadata
        +Shares Data Under Law
    }
    class Citizen {
        +Lacks Awareness
        +No Opt-Out Option
    }
    Government --> Telecom : "Requests Data"
    Telecom --> Citizen : "Tracks Location"

3. Digital Divide: Nepal’s Challenges

Data Comparison (2023):

Solutions Being Tested:

1. Community Wi-Fi Hubs: Nepal Telecom’s rural hotspots in Ilam and Darchula.
2. Subsidized Devices: eSewa’s "Digital Inclusion" program offers discounted tablets.
3. Mobile Money: Khalti and eSewa enable cashless transactions in remote areas.

Worked Example: Bridging the Divide in Sindhupalchowk

• Problem: Only 20% of schools had internet in 2020.
• Solution: World Education installed solar-powered Wi-Fi in 50 schools.
• Impact: Student performance in digital exams improved by 40%.

Exam Tip

This unit is highly conceptual and often tested through:

1. Scenario-Based Questions:

   • "A Nepali bank uses AI to deny loans to rural applicants. Discuss the ethical issues and legal recourse."
   • How to Answer: Use the bias → accountability → transparency framework.

2. Comparison Tables:

   • "Compare the digital divide in Nepal and India, citing two causes and solutions for each."
   • Tip: Use real data (e.g., NTC vs. BSNL penetration rates).

3. Case Study Analysis:

   • "How does Pathao’s use of driver location data raise privacy concerns under Nepal’s Electronic Transactions Act?"
   • Structure:
     • Step 1: Explain surveillance technology (GPS tracking).
     • Step 2: Link to Act’s weaknesses (no data minimization clause).
     • Step 3: Suggest solutions (e.g., anonymization, user consent).

4. Short-Answer Definitions:

   • "Define ‘algorithmic bias’ with a Nepali example."
   • Model Answer: "Algorithmic bias occurs when AI systems produce unfair outcomes due to flawed training data. Example: A Kathmandu-based fintech app’s credit scoring model rejected 25% more rural women applicants because historical data lacked their transaction records."

Visual Cheat Sheet for Exam:

mindmap
  root((Emerging Issues))
    AI
      Bias
      Transparency
      Accountability
    Surveillance
      Privacy vs Security
      Legal Frameworks
    Digital Divide
      Causes
      Solutions
      Nepal Examples