Module overview

Category Content

 What is Artificial Intelligence

(AI)
 Three stages of AI: narrow,
Understanding AI general, super
 Key milestones in the
development of AI

 AI in Everyday Life
Practical Uses of AI  AI in different sectors

 Data Privacy, Security Risks and

Best Practices
 Critical Evaluation
AI Risks, Best Practices and  Ethical and Responsible Use of
Ethical Considerations Generative AI
 Societal and Environmental
Impacts

 Generative AI Tools Overview

Getting Started with Generative AI  Setting Up Generative AI Tools

 Writing Effective Prompts

 Using AI Tools
Using Generative AI Effectively
 Refining and Using Outputs

Comprehensive ICDL Study Notes: Artificial Intelligence (AI)

Module 1: Content Understanding AI
What is Artificial Intelligence (AI)?
Artificial Intelligence is a branch of computer science that builds systems capable of
performing tasks that typically require human intelligence. [1, 2, 3]
┌─────────────────────────────────────┐
│ Artificial Intelligence │
│ Broad field of computing systems │
└──────────────────┬──────────────────┘
▼
 ┌─────────────────────────────────────┐
│ Machine Learning │
│ Systems that learn from data │
└──────────────────┬──────────────────┘
▼
┌─────────────────────────────────────┐
│ Deep Learning │
│ Multi-layer neural networks │
└─────────────────────────────────────┘
Core Concepts
 Machine Learning (ML): A subset of AI where computers learn from data without
explicit programming.
 Deep Learning: A subset of ML using multi-layered artificial neural networks to mimic
the human brain.
 Data Training: Feeding large datasets into algorithms so the system can recognize
patterns. [1, 2, 3, 4, 5]

The Three Stages of AI

AI development is classified into three distinct evolutionary phases based on capability

and adaptability. [1, 2, 3, 4]

1. Artificial Narrow Intelligence (ANI)

 Definition: Also known as "Weak AI."

 Capability: Systems designed to handle a single, highly specific task.
 Limitation: Cannot apply its knowledge to new fields outside its programming.
 Examples: Apple's Siri, Google Translate, chess software, spam filters. [1, 2, 3, 4, 5]
2. Artificial General Intelligence (AGI)
 Definition: Also known as "Strong AI" or "Human-Level AI."
 Capability: A theoretical system possessing the ability to understand, learn, and apply
knowledge across any intellectual task a human can.
 Limitation: Does not yet exist; currently remains a subject of research. [1, 2, 3, 4, 5]
3. Artificial Superintelligence (ASI)
 Definition: AI that surpasses the collective intellect of humankind.
 Capability: Performs across all disciplines, including scientific creativity, general
wisdom, and social skills.
 Limitation: A purely theoretical concept. [1, 2, 3, 4]
 Key Milestones in AI History

The evolution of AI spans decades of theoretical shifts, technical breakthroughs, and

seasonal periods of stagnant funding known as "AI Winters." [1, 2, 3, 4]

1950 1956 1997 2012

2022
│ │ │ │
│
▼ ▼ ▼ ▼
▼
┌──────────────┐ ┌──────────────┐ ┌─────────────────┐
┌───────────────┐ ┌───────────────┐
│ Turing Test │ │ Dartmouth │ │ IBM Deep Blue │ │ AlexNet
│ │ ChatGPT │
│ Proposed │ │ Conference │ │ Defeats │ │ Breakthrough
│ │ Launch │
│ │ │ "AI" Coined │ │ Garry Kasparov │ │ Deep
Learning │ │ Consumer GenAI│
└──────────────┘ └──────────────┘ └─────────────────┘
└───────────────┘ └───────────────┘
 1950: The Turing Test: Alan Turing publishes "Computing Machinery and Intelligence,"
introducing a test for machine intelligence.
 1956: The Dartmouth Workshop: Organized by John McCarthy, where the term
"Artificial Intelligence" is coined.
 1970s–1980s: AI Winters: Periods of reduced funding and interest due to inflated
expectations and technical limitations.
 1997: Deep Blue Defeat: IBM’s Deep Blue supercomputer defeats world chess
champion Garry Kasparov.
 2012: Deep Learning Boom: The AlexNet architecture wins the ImageNet competition,
proving the power of deep neural networks.
 2022: Generative AI Shift: OpenAI releases ChatGPT, bringing conversational
generative models to the mainstream public. [1, 2, 3, 4, 5]

Module 2: Practical Uses of AI

AI in Everyday Life
 AI operates continuously in the background of everyday digital interactions to optimize
user convenience. [1]

 Virtual Assistants: Voice activation platforms convert speech-to-text to set alarms, pull
data, and handle smart appliances.
 Recommendation Engines: E-commerce and streaming algorithms analyze past clicks
to serve matching products, music, or films.
 Smart Navigation: Global positioning software calculates traffic density, routes, and
arrival times dynamically.
 Text Prediction: Keyboards and email clients use language probabilities to correct
spelling and suggest subsequent words. [1, 2, 3, 4, 5]

AI Across Sectors

Industries deploy customized AI structures to solve operational bottlenecks, cut costs,

and increase accuracy. [1, 2, 3, 4]

Sector [1, 2, 3,
Primary Applications Real-World Benefit
4, 5]
Image scanning, predictive diagnostics, Early tumor discovery, accelerated
Healthcare
molecule pairing. drug development.
Pattern alerts, high-frequency trading, Immediate fraud identification, market
Finance
automated credit risk. optimization.
Modular curricula, automatic grade queues, Personalized learning speed, reduced
Education
vocabulary bots. teacher workload.
Dynamic warehouse tracking, predictive Lower inventory overhead, 24/7 client
Retail
supply chains, chatbots. resolution.

Module 3: AI Risks, Best Practices, and Ethical Considerations

Data Privacy and Security Risks

The dependence of AI on massive quantities of information creates technical

vulnerabilities for individuals and institutions. [1, 2, 3]
 Risks

 Data Leakage: Inputting proprietary or private customer information into public AI

models makes that data part of the public pool.
 AI-Powered Cyberthreats: Malicious actors utilize AI to craft highly convincing
phishing emails or automate security exploits.
 Deepfakes: Synthetic voice and video generation used to execute identity fraud and
corporate impersonation. [1, 2]
Best Practices
 Data Minimization: Exclude names, addresses, and corporate secrets from system
prompts.
 Anonymization: Strip all personally identifiable information (PII) before loading files into
analytical models.
 Access Control: Enforce multi-factor authentication (MFA) and strict access logging on
corporate AI dashboards. [1, 2, 3]

Critical Evaluation of AI Outputs

AI applications operate on probability, meaning their responses require human

verification before operational use. [1, 2]

 Hallucinations: Instances where an AI system generates false information, fake

citations, or fabricated statistics confidently.
 Data Bias: Models trained on historically skewed data reproduce and amplify those
same prejudices in hiring, lending, or legal scoring.
 Mitigation Strategy: Establish a "Human-in-the-Loop" workflow where qualified
professionals verify every automated claim against authoritative reference materials. [1,
2, 3, 4, 5]

Ethical and Responsible Use of Generative AI

Using generative tools requires adherence to intellectual property regulations and

academic honesty principles. [1, 2, 3]
 Copyright Infringement: AI models are often trained on copyrighted books, code, and
artwork without clear consent or compensation.
 Plagiarism: Copying raw AI outputs and submitting them as human-created work
breaches educational and professional standards.
 Transparency Mandate: Best practices demand explicit disclosures (e.g., "Drafted with
assistance from AI") whenever public material relies on automated generation. [1, 2, 3,
4, 5]

Societal and Environmental Impacts

The widespread deployment of AI reshapes labor markets and strains global ecological
infrastructure. [1, 2]

 Job Displacement: Routine data processing, administrative roles, and structural

customer service tasks face high automation risks. [1, 2, 3, 4, 5]
 Energy Consumption: Training complex models and running massive data centers
demands substantial electricity, magnifying carbon footprints. [1, 2, 3, 4, 5]
 Electronic Waste (E-waste): The intensive processing requirements of advanced AI
accelerate the obsolescence of older computer hardware. [1]

Module 4: Getting Started with Generative AI [1]

Generative AI Tools Overview

Generative AI refers to models that produce completely new text, images, code, or
audio based on historical training data. [1, 2, 3, 4, 5]

┌───────────────────────────────┐
│ Generative AI Tools │
└───────────────┬───────────────┘
│
┌──────────────────────────┼──────────────────────────┐
▼ ▼ ▼
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
 │ Text Generators │ │Image Generators │ │ Code Assistants │
│ ChatGPT, Claude,│ │ Midjourney, │ │ GitHub Copilot, │
│ Gemini │ │ DALL-E, Stable │ │ Tabnine │
└─────────────────┘ └─────────────────┘ └─────────────────┘
 Text Engines (Large Language Models): OpenAI ChatGPT, Anthropic Claude,
Google Gemini.
 Visual Synthesis Engines: Midjourney, DALL-E 3, Stable Diffusion.
 Development Code Assistants: GitHub Copilot, Amazon Q, Tabnine. [1, 2, 3, 4, 5]

Setting Up Generative AI Tools

1. Account Registration: Sign up using certified credentials or via corporate single sign-
on (SSO) interfaces.
2. Interface Navigation: Locating the prompt field, history sidebar, model selection toggle,
and account settings panel.
3. API (Application Programming Interface) Architecture: Constructing secure
developer keys to run AI logic directly inside third-party programs. [1, 2, 3, 4, 5]

Module 5: Using Generative AI Effectively

Writing Effective Prompts (Prompt Engineering)

A prompt is the textual instruction passed to an AI model. Clear prompts reduce

hallucinations and deliver focused outputs. [1, 2, 3]

Framework Elements

 Role/Persona: Assign a specific professional identity. (e.g., "Act as an expert data

analyst.")
 Context/Task: Define the background scenario and target goal. (e.g., "Review this
feedback list to uncover why clients are canceling subscriptions.")
 Constraints/Format: Establish boundaries, style, length, or output layout. (e.g.,
"Present the findings in a 3-column markdown table. Do not use industry jargon.") [1, 2,
3, 4, 5]
 Core Applications of AI Tools

 Ideation & Brainstorming: Generating initial marketing headlines, project outlines, or

content themes.
 Drafting Content: Outlining basic corporate correspondence, policy memos, or
structural email templates.
 Summarization: Extracting core takeaways, technical action items, or data themes
from lengthy PDF reports. [1, 2, 3, 4, 5]

Refining and Using Outputs

1. Iterative Prompting: If the first response misses details, feed constructive corrections
back into the prompt thread.
2. Fact Verification: Isolate dates, statistics, legal regulations, and URLs to confirm their
accuracy manually.
3. Tone Adaptation: Rewrite structural AI drafts to inject personal brand style, cultural
nuances, and genuine voice. [1, 2, 3, 4, 5]