CORE AI CONCEPTS & TERMINOLOGY

What is AI?

AI refers to technologies that enable computers to perceive, learn, reason, and assist in human-like decision-making.

How does the Organisation for Economic Co-operation and Development (OECD) define AI?

As a machine-based system using human-defined objectives to produce predictions, recommendations, or decisions influencing real or virtual environments.

How has AI complexity grown over time?

AI parameters grew from ~100 million in 2018 to over 1 trillion by 2022, driven by deep learning advancements and GPU technology. Compute requirements doubled every six months during the deep learning era starting in 2010.

What is machine learning and how does it differ from deep learning?

ML allows computers to learn patterns or reach goals without explicit programming. Deep learning uses large datasets and neural networks inspired by the human brain to improve pattern recognition and reduce prediction error.

What is the difference between supervised and unsupervised learning?

Supervised learning trains AI on labelled datasets, while unsupervised learning exposes AI to unlabelled data to find solutions autonomously.

What is machine perception in AI?

The ability of AI systems to recognize, analyze, and respond to data—allowing them to "see," "hear," "comprehend," and "reason." NLP is often a primary interface for perception.

How does AI systems control machines?

Robotics combines machine learning and perception to automate movement and interaction in environments, following a "sensing → planning → acting" cycle.

What are expert systems in AI?

AI systems that emulate human decision-making by applying rules (inference engine) to a knowledge base of facts and rules.