Module 4 of 5
What AI is and is not. How AI is already in your life. Using AI tools for learning, writing, and problem-solving.
Fatuma teaches at a secondary school in Mombasa. She has heard 'AI' discussed everywhere — in staff meetings, on the radio, in parent conversations. She has been told AI will replace teachers. She has been told AI will save education. She has been shown a chatbot that wrote a mediocre essay in seconds.
She does not know what to believe — partly because most of what she hears is either alarm or hype, with very little actual explanation of what AI is, how it works, or what it means specifically for her work and her students' futures.
This module is the honest explanation she has been waiting for.
Artificial intelligence is software that has been trained to perform tasks that previously required human intelligence — tasks like recognizing speech, translating between languages, identifying objects in photos, recommending videos, and generating text.
AI systems learn from data. A speech recognition system was trained on millions of hours of recorded human speech until it became good at converting audio to text. An image recognition system was trained on millions of labeled photos until it learned to identify cats, cars, and faces. A language model like ChatGPT or Claude was trained on vast amounts of text from the internet, books, and other sources until it became good at predicting and generating coherent language.
The key insight: AI does not think, understand, or feel. It recognizes patterns in the data it was trained on and generates outputs based on those patterns. When ChatGPT writes an essay, it is not 'thinking about' the topic — it is generating text that statistically resembles the kinds of essays it encountered during training. This distinction matters, because it tells you both what AI can do well and where it will fail.
AI is very good at:
Language tasks — drafting text, translating, summarizing, editing, answering questions based on its training
AI is poor at:
Knowing when it does not know something — AI systems frequently generate confident-sounding incorrect information
Current events — most AI models have a knowledge cutoff date and do not know what has happened recently
Cultural nuance and local context — an AI trained primarily on English-language internet content knows less about Kakamega than about California
The African AI market is projected to grow at 40% annually through 2030, reaching $8 billion. However, African languages represent less than 1% of the data used to train leading AI models — meaning AI tools work less reliably in Swahili, Amharic, Yoruba, Hausa, and other African languages than in English or French.
Source: African Development Bank — Digital Transformation Strategy 2021–2025; Masakhane Research Community — African Languages in AI Report (2023)
You are already using AI every day — you may simply not have recognized it as such.
M-Pesa fraud detection: When M-Pesa flags an unusual transaction, an AI system made that decision — analyzing patterns in your transaction history and comparing them to known fraud signatures.
Google Maps routing: When Google Maps calculates the fastest route from Westlands to Karen in real-time traffic, AI is processing thousands of data points from other drivers' GPS signals to predict congestion.
YouTube recommendations: The videos that appear in your suggested feed are selected by an AI recommendation engine trained on billions of user behaviors.
Google Translate: When you translate a WhatsApp message from English to Swahili, a neural machine translation AI — not a human — is producing that translation.
Safaricom network optimization: The network capacity allocation that determines whether your call drops or connects uses AI to predict demand patterns across towers.
AI is not coming. It is here. The question is whether you will interact with it passively — as a user of systems someone else designed — or actively — as someone who knows how to direct it toward your specific goals.
Three AI tools are free, available in Kenya, and useful for education, business, and personal productivity today:
ChatGPT (chat.openai.com) is built by OpenAI and is the most widely used AI writing and question-answering tool in the world. The free version (GPT-3.5) is powerful enough for most everyday tasks. You access it via browser on your phone — no app download required. It is available in English and has limited Swahili capability.
Claude (claude.ai) is built by Anthropic and is known for careful, detailed reasoning and more cautious handling of complex topics. The free tier is useful for drafting, analysis, and research questions. It works well for professional writing tasks — letters, proposals, business plans.
Google Gemini (gemini.google.com) is Google's AI assistant, integrated with Google's search and productivity tools. It has stronger Swahili capabilities than ChatGPT and integrates with Gmail and Google Docs, which is useful if your business already uses Google tools.
How to use these tools effectively:
Be specific. Instead of 'write me a business plan,' write: 'Write a one-page business plan for a second-hand furniture refurbishing business in Thika, Kenya, targeting middle-income customers. Include a market overview, revenue model, and marketing approach.'
Give context. AI produces better outputs when it knows who you are, what you need, and who the audience is.
Verify everything. AI will occasionally state incorrect information with confidence. Any factual claim that matters — prices, laws, medical information — verify through an official source.
Iterate. If the first output is not right, say 'That is good but too formal — make it more conversational' or 'Add more detail to the financial section.' AI tools improve with direction.
Lelapa AI is an African-founded AI research company focused on building language AI tools for African languages — beginning with Zulu, Sesotho, and isiXhosa, and expanding toward Swahili and other East African languages.
The problem they are solving: leading AI models were trained almost entirely on English, French, and Mandarin content. When a Zulu speaker uses ChatGPT in their home language, they are interacting with a system that knows Zulu as a small appendix to its English-language core. The results are often inaccurate, culturally misaligned, and less useful than the same tool used in English.
Lelapa AI's approach: create datasets of African language text, fine-tune existing models on that data, and develop AI tools that understand African cultural context — not just linguistic translation.
Masakhane, a pan-African research community with over 700 members across the continent, takes a complementary approach: crowdsourcing the creation of African language datasets through volunteer linguists and researchers.
The implication for IHA Advance learners: the AI tools available today are less useful in Swahili than in English. This is a known limitation that African researchers and technologists are actively working to close. By the time you are using AI tools professionally in 5 years, Swahili-native AI will likely be significantly more capable. The investment in AI literacy now prepares you to use those tools when they arrive.
The honest answer to 'will AI take my job?' is: it depends on what your job actually involves.
AI is most likely to automate tasks that are: repetitive and rule-based; involve processing structured information; require little contextual judgment or physical presence. Examples include: basic data entry, standard document drafting, routine customer service responses, and some forms of image classification.
AI is least likely to replace work that involves: physical presence and manual dexterity (construction, farming, healthcare delivery); complex human relationships and trust (community health work, teaching, counseling, ministry); creative and contextual judgment (business strategy, community organizing, artisanal craft); and adaptation to unpredictable local conditions (informal market trading, smallholder agriculture).
For African economies, the more important question may not be 'will AI replace jobs' but 'will Africans be able to use AI to create new value and access new markets.' The evidence so far suggests yes — but the people best positioned to capture that value are those who understand how to use AI tools, not those who avoid them.
A World Economic Forum analysis projects that AI will displace 85 million jobs globally by 2025 while creating 97 million new roles — a net positive. However, new roles require different skills, and the transition will be uneven. In Africa, the greater near-term risk is not job displacement but job exclusion — skilled Africans unable to compete in global markets because they lack AI fluency.
Source: World Economic Forum — Future of Jobs Report 2023; McKinsey Global Institute — The Future of Work in Africa (2021)
Go to chat.openai.com or claude.ai on your phone browser. Create a free account.
You do not need to download an app — both work in the browser. Use your Gmail or email address to register. This takes 3 minutes. If you have data limitations, do this over Wi-Fi.
Give it a specific, real task from your life or work.
Examples: 'Write a professional text message I can send to a supplier asking for a 30-day payment extension, in a respectful tone.' Or: 'I run a small catering business in Nakuru. Give me 5 marketing messages I can use on WhatsApp to attract customers for Christmas.' Or: 'Explain to me in simple terms what I need to do to register a business in Kenya.' Use the output — edit it, send it, or act on it. The goal is one real output, not just exploration.
Write 3 sentences describing what you found: what worked, what surprised you, and one limitation you noticed.
Keep this written record. Over the coming months, as you use AI tools more, you will notice your prompts getting more specific and your outputs getting more useful. That progression is a skill being developed — not a magic tool being discovered.
AI is a tool — one of the most powerful tools available to small business owners, educators, and community workers in Africa right now — and it is free. The advantage goes to those who learn to use it deliberately, not those who wait to see what happens.
Want to go further? These free resources are the next step:
ChatGPT — OpenAI's free AI assistant — Start here for writing, research, and question-answering tasks chat.openai.com
Masakhane — African AI Research Community — Learn about the work being done to make AI work better in African languages masakhane.io
Google's AI Essentials Course — Free 5-hour course on understanding and using AI tools — available via Coursera coursera.org/learn/google-ai-essentials
Answer this question before completing the module
Describe one task in your work, studies, or daily life that you could automate or improve with an AI tool you learned about in this module. Write out exactly how you would use it: what input you would give, what output you would expect, and how it would save you time or improve your results. Then try it and write two sentences about what happened.
Score 2 out of 3 to complete this module
1. You ask an AI assistant to help you write a business proposal. It produces a well-written draft but includes a claim that is factually incorrect. What does this illustrate about current AI tools?
2. Machine learning is a type of AI. Which of the following best describes how it works?
3. Which of the following is a practical example of AI being used in everyday African life today?