Understanding AI
Before the framework, the guide sets out what AI actually is — and adopts an international definition rather than writing its own.
The definition the Ministry adopts
“A machine-based system designed to operate with varying levels of autonomy; that may exhibit adaptiveness after deployment, whether explicitly or implicitly; and that infers, from the input it receives, how to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments.”
— OECD, 2024: Regulation (EU) 2024/1689p.15
How AI learns
Data
Text, images, numbers, audio or video used as input for the machine to “learn”.p.17
Algorithm
The set of instructions or mathematical formulas that are the heartbeat of AI — telling the machine how to process, interpret and learn from data.p.17
Learning method
Based on the data type, the machine learns from existing or current data to predict, decide, classify or create new content.p.17
Machine Learning
The main subset of AI; applies algorithms to data. Outputs: predictions, classifications, reports, diagnoses, clustering.
Deep Learning
A more specialised subset using artificial neural networks for complex problems such as understanding human language or recognising objects in video.
Generative AI
Typically powered by large language models trained on huge volumes of data. Named examples in the guide: ChatGPT and Gemini AI.
Four eras
- 1950s
Birth
The term “Artificial Intelligence” is coined, marking the start of the aspiration to build machines that mimic human thinking.p.15
- 1970s–80s
AI Winter
Progress slows because of technological limits and lack of funding.p.15
- 1990s–2000s
Resurgence
Computing power rises; Deep Blue defeats the world chess champion.p.15
- 2010s–now
Modern Era
Deep learning explodes and generative AI enters everyday life.p.15
Where AI already appears
(i) Virtual assistants
Google Gemini, Siri
(ii) Recommendation systems
Netflix, Shopee
(iii) Automatic translation
Google Translate, Microsoft Translator
(iv) Facial recognition
Face ID, Face Unlock
(v) Customer-service chatbots
Text or voice interaction
The examples the guide itself names · p.18
Four types relevant in education p.19
Adaptive learning platforms
Dynamically adjust content and difficulty to each student's needs.
Educational chatbots
Interact with students, give learning support, answer common questions.
Learning analytics
Analyse student data to spot learning patterns, predict students at risk of dropping out, and plan early intervention.
Content generators
Create new material — text, images, music — from user prompts.