AI, Surveillance & Digital Divide: Ethics, Risks & Global Impact

Key Concepts & Definitions

1. Artificial Intelligence (AI) and Ethical Dilemmas

Definition: AI refers to machines or software that mimic human intelligence (learning, reasoning, problem-solving). Ethical AI ensures fairness, transparency, and accountability in its design and deployment.

Core Ethical Challenges:

• Bias & Discrimination: AI trained on biased data (e.g., facial recognition failing for darker skin tones) perpetuates inequality.
• Autonomy vs. Control: Should AI make life-altering decisions (e.g., loan approvals, medical diagnoses) without human oversight?
• Job Displacement: Automation threatens roles in manufacturing, customer service, and even creative fields (e.g., AI-generated art replacing illustrators).

How It Works: AI systems rely on algorithms (step-by-step logic) and machine learning (self-improving models). For example:

• Supervised Learning: Trained on labeled data (e.g., spam emails marked "spam" or "not spam").
• Unsupervised Learning: Finds hidden patterns (e.g., Netflix recommending shows based on user behavior).
• Reinforcement Learning: Learns by trial and error (e.g., AlphaGo beating human Go champions).

2. Surveillance: Scope, Tools, and Ethical Boundaries

Definition: Surveillance is the systematic monitoring of individuals or groups, often using technology (CCTV, facial recognition, data tracking). Mass surveillance targets entire populations without suspicion.

Tools of Surveillance:

Real-World Example:

• Nepal’s Traffic Management System: Cameras and AI analyze traffic flow in Kathmandu to reduce congestion. While efficient, critics argue it could enable predictive policing (targeting "suspicious" areas based on data).

3. Digital Divide: Causes and Consequences

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

• Economic inequality (e.g., rural vs. urban Nepal).
• Infrastructure gaps (e.g., limited internet in remote areas).
• Digital literacy (e.g., elderly or low-income groups excluded from online services).

Types of Digital Divide:

mindmap
  root((Digital Divide))
    Global-South vs. Global-North
      Example: Nepal (slow internet) vs. South Korea (5G everywhere)
    Urban vs. Rural
      Example: Kathmandu’s fiber optics vs. Sindhupalchok’s dial-up
    Demographic
      Example: Youth using WhatsApp vs. elders relying on landlines
    Access vs. Usage
      Example: Having a smartphone but not knowing how to use it

Impact on Society:

• Education: Online learning platforms (e.g., Swayam Prabha in Nepal) exclude students without devices.
• Healthcare: Telemedicine fails in areas with poor connectivity (e.g., Nepal Health Research Council’s remote clinics).
• Economic Opportunities: E-commerce (e.g., Daraz) favors urban entrepreneurs over rural artisans.

Emerging Issues in Depth

1. AI in Nepal: Opportunities and Risks

Case Study: AI in Banking (Nabil Bank, Global IME)

• Use: AI chatbots (e.g., Nabil Bank’s "Nabil Assist") handle customer queries 24/7.
• Risk: If the AI misinterprets loan applications, it could deny credit unfairly (e.g., rejecting a small business owner due to biased data).
• Ethical Fix: Banks must audit AI decisions for fairness (e.g., Algorithmic Transparency Laws proposed in the EU).

Worked Example: Loan Approval Bias Suppose an AI model trained on historical data rejects 80% of loan applications from women because past data showed fewer female borrowers. This is discriminatory by design. Solution: Use diverse training data and human oversight in high-stakes decisions.

2. Surveillance in Nepal: Laws and Controversies

Key Laws:

• Electronic Transactions Act (2008): Regulates digital transactions but lacks strong privacy protections.
• Right to Privacy (2075): Recognizes privacy but doesn’t ban mass surveillance.
• Data Localization Rules (2018): Requires sensitive data (e.g., NID, bank records) to be stored in Nepal.

Controversial Cases:

• Nepal Police’s Facial Recognition Trial (2022): Tested in Thamel to "prevent crime," but critics say it’s overreach without public consent.
• Chinese Tech in Nepal: Companies like Huawei and ZTE have sold surveillance tech to Nepal Police, raising concerns about data sovereignty (who controls the data?).

3. Digital Divide in Nepal: A Closer Look

Data Highlights (2023):

Government Initiatives:

• Digital Nepal Project (2018): Aims to connect all villages by 2025 (currently at 60%).
• Free Wi-Fi in Public Spaces: Trials in Pokhara and Bhaktapur, but limited coverage.
• E-Sewa Expansion: More services (e.g., citizen ID verification) but excludes offline users.

Worked Example: Rural Farmer’s Struggle A farmer in Dolakha cannot use e-Kisan (agricultural app) because:

1. No smartphone.
2. Poor 3G signal.
3. Doesn’t know how to register. Solution: Community digital hubs (like Nepal Telecom’s "Digital Seva Kendra") could bridge the gap.

Ethical Frameworks for Emerging Tech

In the Real World

1. WhatsApp’s End-to-End Encryption (AI + Privacy)

   • How it uses AI: Detects spam/bullying via NLP (Natural Language Processing).
   • Ethical Conflict: Balances security (stopping abuse) with privacy (governments want access to messages).
   • Nepal Link: Nepal Police has demanded WhatsApp decryption keys for investigations, raising surveillance vs. privacy debates.

2. Daraz’s AI Recommendation System

   • How it works: Uses collaborative filtering (like Amazon) to suggest products.
   • Risk: If the AI over-recommends expensive items, it could exploit low-income users.
   • Ethical Fix: Daraz could cap price suggestions for vulnerable groups.

3. NTC’s Fiber Optic Expansion (Digital Divide)

   • Goal: Reduce urban-rural gap by laying fiber in 100 districts.
   • Challenge: Corruption in tenders and high costs delay projects.
   • Real Impact: In Jumla, students now access online TU exams, but electricity cuts still disrupt learning.

Exam Tip

How This Unit is Tested:

1. Case Study Analysis (30%):

   • Expect short-answer questions on:
     • "How does facial recognition in Nepal violate the Right to Privacy Act?"
     • "Design an ethical AI policy for a Nepali bank."
   • Tip: Use PESTLE analysis (Political, Economic, Social, Technological, Legal, Environmental) to structure answers.

2. Scenario-Based Questions (25%):

   • Example:

     "A rural school in Kavrepalanchok wants to use AI tutors but lacks electricity. What are the ethical and technical challenges?"

   • Answer Framework:
     • Ethical: Digital divide, accessibility.
     • Technical: Offline AI tools, solar-powered solutions.
     • Legal: Compliance with Nepal’s Education Act.

3. Comparison Tables (20%):

   • Likely to compare:
     • AI in Nepal vs. China (e.g., social credit vs. loan approvals).
     • Digital Divide solutions (e.g., government vs. NGO approaches).

4. Long Answer (25%):

   • Sample Question:

     "Discuss the ethical implications of mass surveillance in Nepal, with reference to the Electronic Transactions Act and global examples."

   • Structure:
     1. Define mass surveillance (tools, scope).
     2. Nepal’s legal framework (weaknesses in ETA 2008).
     3. Global case study (e.g., China’s social credit system).
     4. Ethical theories (deontology vs. utilitarianism).
     5. Recommendations (e.g., Data Protection Bill 2076).

Key Formulas/Terms to Memorize:

• Algorithmic Bias Formula: $\text{Bias}=\frac{\text{Error Rate (Minority Group)}-\text{Error Rate (Majority Group)}}{\text{Average Error Rate}}$
• Digital Divide Index (DDI): $\text{DDI}=\frac{\text{Urban Access}-\text{Rural Access}}{\text{Total Population}}\times 100$

Visual Cheat Sheet for Exam:

flowchart TD
    A["AI Ethics"] --> B["Bias<br/>Discrimination"]
    A --> C["Autonomy<br/>Accountability"]
    A --> D["Job Displacement"]
    B --> E["Nepal Bank Loan AI"]
    C --> F["EU AI Act<br/>Nepal Draft Laws"]
    D --> G["Reskilling Programs<br/>NTA Vocational Training"]
    H["Surveillance"] --> I["Facial Recognition<br/>Nepal Police"]
    H --> J["Data Privacy<br/>Right to Privacy Act"]
    K["Digital Divide"] --> L["Urban-Rural Gap<br/>NTC Fiber Project"]
    K --> M["Demographic Gap<br/>Elderly Exclusion"]