Business School – Understanding AI and how it Works – Lesson 2

Understanding AI Subfields

This AI lesson provides detailed breakdowns of:

• Machine Learning – The foundation of AI with African applications in agriculture, credit scoring, and healthcare
• Deep Learning – Advanced neural networks used in medical imaging and wildlife conservation
• Natural Language Processing – Language understanding tech (with discussion of the African language challenge)
• Computer Vision – Visual recognition for agriculture, traffic management, and mobile banking

Each section includes how it works and real-world African applications.

Introduction

Artificial Intelligence is not a single technology but rather a collection of specialised fields, each addressing different aspects of how machines can learn, understand, and interact with the world. This tutorial breaks down the four key AI subfields that are driving innovation across Africa and globally.

Part 1: The Four Key AI Subfields

1. Machine Learning (ML)

What It Is: Machine Learning is the foundation of modern AI. It’s the science of teaching computers to learn from data and improve their performance over time without being explicitly programmed for every task. Instead of following rigid rules, ML systems identify patterns in data and make predictions or decisions based on what they’ve learned.

How It Works: Imagine teaching a child to recognize fruits. You don’t give them a rulebook; you show them many examples of apples, oranges, and bananas until they can identify each fruit on their own. Machine Learning works similarly—algorithms analyze thousands or millions of examples to learn patterns and make accurate predictions on new, unseen data.

Real-World Applications in Africa:

• Agriculture: Farmers in Kenya use ML-powered apps to predict crop diseases by analyzing photos of plant leaves, helping prevent widespread crop failure
• Credit Scoring: In Nigeria and South Africa, fintech companies use ML to assess creditworthiness for people without traditional banking histories, expanding financial inclusion
• Weather Forecasting: ML models analyze historical weather patterns to provide more accurate forecasts for farmers planning planting seasons
• Healthcare: ML algorithms help diagnose diseases like tuberculosis and malaria from medical images, particularly useful in areas with few doctors
• Telecommunications: Mobile networks use ML to predict network failures and optimize coverage in urban and rural areas

2. Deep Learning (DL)

What It Is: Deep Learning is a specialized subset of Machine Learning inspired by how the human brain works. It uses artificial neural networks with many layers (hence “deep”) to process information. Think of it as Machine Learning’s more sophisticated cousin—capable of handling incredibly complex tasks that traditional ML struggles with.

How It Works: Deep Learning models consist of interconnected layers of artificial neurons. Each layer processes information and passes it to the next, gradually building understanding from simple features to complex concepts. For example, when recognizing a face, early layers might detect edges, middle layers identify facial features like eyes and noses, and final layers recognize specific individuals.

Real-World Applications in Africa:

• Medical Imaging: Hospitals in Ghana and Rwanda use deep learning to analyze X-rays and CT scans, detecting cancers and other conditions even when radiologists are scarce
• Satellite Image Analysis: Deep learning helps monitor deforestation in the Congo Basin, track urban growth, and assess agricultural land use across the continent
• Speech Recognition: Systems are being developed to understand African languages and accents, improving voice assistants for local populations
• Wildlife Conservation: Camera traps use deep learning to automatically identify and count endangered species like elephants and rhinos, supporting anti-poaching efforts
• Autonomous Vehicles: Research in South Africa explores self-driving technology adapted for African road conditions

3. Natural Language Processing (NLP)

What It Is: Natural Language Processing enables computers to understand, interpret, and generate human language. It bridges the gap between human communication and computer understanding, allowing machines to read text, hear speech, interpret meaning, measure sentiment, and even generate coherent responses.

How It Works: NLP combines linguistic rules, statistical methods, and machine learning to break down language into components computers can process. It handles challenges like context, ambiguity, slang, and cultural nuances. Modern NLP uses techniques like tokenization (breaking text into words), sentiment analysis (determining emotional tone), and transformer models (understanding context across long passages).

Real-World Applications in Africa:

• Customer Service Chatbots: Banks and telecoms across Africa use NLP-powered chatbots to handle customer queries in multiple languages, providing 24/7 support
• Translation Services: NLP helps translate between African languages and global languages, breaking down communication barriers for business and education
• News Monitoring: Media organizations use NLP to track breaking news, analyze public sentiment, and identify trending topics across social media
• Legal Tech: In South Africa, NLP tools help lawyers search through thousands of legal documents quickly, making legal services more accessible
• Education: Language learning apps use NLP to help students practice reading and writing in both local and international languages

The African Language Challenge: A major limitation of current NLP systems is their poor performance with African languages. Research shows that GPT-4 recognizes sentences in Hausa only 10-20% of the time, highlighting the need for locally developed AI solutions that truly understand Africa’s linguistic diversity.

4. Computer Vision (CV)

What It Is: Computer Vision gives machines the ability to “see” and understand visual information from the world—images, videos, and real-time camera feeds. It’s about teaching computers to extract meaningful information from visual data, much like human vision allows us to navigate, recognize objects, and interpret our surroundings.

How It Works: Computer Vision systems use cameras or image sensors to capture visual data, then apply algorithms to identify patterns, objects, faces, movements, and scenes. Modern CV heavily relies on deep learning, using neural networks trained on millions of images to recognize everything from simple shapes to complex scenarios.

Real-World Applications in Africa:

• Agricultural Monitoring: Drones equipped with computer vision fly over farms in Ethiopia and Kenya, detecting pest infestations, measuring crop health, and estimating yields before harvest
• Healthcare Diagnostics: CV systems analyze medical images to detect diabetic retinopathy in eye scans and identify skin conditions, particularly valuable in rural clinics without specialists
• Traffic Management: Cities like Lagos and Nairobi use computer vision to monitor traffic flow, detect accidents, and optimize traffic light timing during peak hours
• Retail Analytics: Stores use CV to track customer movements, analyze shopping behavior, and prevent theft through automated surveillance
• Mobile Money Security: Banks use facial recognition powered by computer vision to verify identities for mobile banking, reducing fraud while improving accessibility
• Manufacturing Quality Control: Factories use CV to inspect products on assembly lines, identifying defects faster and more consistently than human inspectors

How These Subfields Work Together

These four AI subfields don’t operate in isolation—they often combine to create powerful solutions:

Example: A Smart Healthcare App

• Computer Vision analyzes a photo of a skin lesion
• Deep Learning processes the complex image patterns to detect potential melanoma
• Machine Learning compares the case to thousands of previous diagnoses to estimate risk
• Natural Language Processing generates a report in the patient’s preferred language and communicates findings to healthcare providers

Example: An Agricultural Assistant

• Computer Vision identifies crops and pests from drone imagery
• Machine Learning predicts optimal planting times based on weather and soil data
• Natural Language Processing provides farmers with advice in their local language via text or voice
• Deep Learning continuously improves predictions as more farming data becomes available

Why This Matters for Africa

Understanding these AI subfields is crucial because:

1. Economic Opportunity: AI could contribute $2.9 to $4.8 billion to Africa’s economy by 2030
2. Local Solutions: Africa needs AI systems built for African contexts, languages, and challenges
3. Digital Sovereignty: Understanding AI helps African developers build solutions rather than just importing foreign technology
4. Job Creation: As AI adoption grows, demand for professionals skilled in these areas will surge
5. Problem Solving: These technologies can address uniquely African challenges in agriculture, healthcare, education, and infrastructure

Key Takeaway

Machine Learning, Deep Learning, Natural Language Processing, and Computer Vision represent different approaches to making machines intelligent. Each has unique strengths, and together they’re transforming industries across Africa—from helping farmers maximize yields to improving healthcare access in remote areas. The challenge now is ensuring these technologies are developed with African needs, languages, and contexts in mind.

Check out our Lesson 1 on AI: – How AI works and what the different approaches are.