BinaryTree

Digital Literacy & Digital Entrepreneurship

BinaryTree Program – Senegal

Digital Literacy and Digital Entrepreneurship

BinaryTree

Week 4

AI Usage and Key Concepts

Last Week's Recap

• Last week, we explored creating a digital presence via LinkedIn, Instagram, Gmail, and Canva — with an emphasis on design, communication, and professionalism.

• These platforms can now be enhanced with AI to save time, improve quality, and achieve your goals faster.

• This week, we connect these tools to intelligent systems that improve their usefulness in academic and professional tasks.

Q&A

Do you have any questions about last week?

​

Ask them in the chat now!

What is AI?

• AI (Artificial Intelligence) is a branch of computer science that creates systems capable of performing tasks that normally require human intelligence.

These systems can:

• Learn from experience
• Adapt to new data
• Perform tasks such as speech recognition, decision making, and text translation

AI is already used in healthcare, finance, education, and entertainment.

What are LLMs?

• Large Language Models (LLMs) are AI systems trained on datasets containing hundreds of billions of words to understand and generate human-like text.

• They can generate articles, answer complex questions, simulate conversations, write code, and perform translations.

• These models are the foundation of tools like ChatGPT, DeepSeek, and Google Gemini.

Activity

4 min

List some applications of AI that you know today.

​

Share your ideas in the chat!

How LLMs Learn

• LLMs are trained using 'deep learning,' which uses neural networks inspired by the human brain.
• These networks process huge amounts of text data to detect patterns in language usage and context.
• During training, the model predicts the next word in a sentence, adjusting its internal parameters to reduce prediction errors. This is repeated billions of times.
• As the model encounters more diverse examples, its ability to understand linguistic context improves — enabling consistent, context-appropriate responses.

Neural Networks and Transformers

• Modern LLMs use a Transformer architecture, which allows the model to assess the relative importance of words to each other — greatly improving contextual understanding.
• Transformers use 'attention mechanisms' to focus on relevant parts of input text, even if words are far apart in a sentence.
• This approach outperforms older models like RNNs, allowing LLMs to process longer texts and generate more natural language.

Practice Question

3–4 min

What is the basic inspiration behind LLMs, and what is their source of inspiration?

​

Discuss in the chat.

Size and Diversity of Datasets

• LLMs are trained on datasets including books, websites, articles, and other written content — totaling hundreds of billions of words. This variety helps the model learn different writing styles, topics, and languages.

• The larger and more diverse the dataset, the better the model generalizes and provides accurate answers across different domains.

• However, biases in training data can influence model outputs, making ethical considerations and data curation important areas of research.

Concrete Applications of LLMs

Daily Life

Powers virtual assistants like ChatGPT, automated customer support, language translation, and content creation tools.

Education

Generates personalized learning materials and instantly responds to student questions, improving global access to knowledge.

Business

Analyzes large amounts of data, summarizes reports, and even writes emails — saving time and improving decision-making.

Using ChatGPT

• ChatGPT can help with a variety of tasks: writing essays, summarizing articles, explaining complex concepts, or preparing for exams.

• It works like a helpful assistant — but you need to check its answers and guide the conversation. AI is only as smart as your questions.

• It is important to use AI to support your learning, not to replace your own effort or understanding.

Using ChatGPT – Good Prompting

To get the most out of ChatGPT, keep your requests concise and specific.

Good Prompt

"Show me what's happening in the healthcare industry right now. Include statistics on patient numbers, patient satisfaction, and success rates."

Bad Prompt

"What's going on with hospitals right now?"

Limitations of ChatGPT

• ChatGPT is not always accurate and may provide incorrect or outdated information.

• It has no personal experiences or emotions — it simply predicts words based on patterns in its training data.

• Users should verify important information and not rely on it for sensitive tasks.

ChatGPT in Numbers

100M+

Monthly users

570 GB

Text data trained on

7–10 min

Avg. session length

Using Pictory for Learning

• Visual learners retain up to 80% more information with video content, making Pictory a powerful teaching tool for reviewing or presenting material.

• Students can transform essays, reports, or presentations into video formats, reinforcing understanding through storytelling and visuals.

• It also stimulates creativity — allowing you to choose themes, music, and pacing to express ideas in a dynamic way.

AI for Image Generation

• Platforms like DALL·E and Midjourney use AI to create detailed, creative images from text descriptions — transforming art creation and marketing design.

• These tools use generative adversarial networks (GANs) or diffusion models to produce high-resolution images for branding, advertising, or storytelling.

• By 2025, the AI-generated art market was expected to grow by more than 20% per year — disrupting traditional creative industries.

Different Types of AI

CNN

Convolutional Neural Networks — used for image recognition, detecting objects in photos or diagnosing diseases from scans. Example: Photovision.

LLM

Large Language Models — understand and generate human text for writing, Q&A, and summarization. Example: ChatGPT, Gemini.

RNN / GAN

RNNs handle sequence tasks (e.g., Siri/Alexa). GANs generate realistic data like synthetic images or deepfakes. Example: NVIDIA GauGAN.