I85PREP
⌘ K
WBCHSE Applied AI study materials
WBCHSE Applied AI study resources

WBCHSE Applied AI: Introduction to Artificial Intelligence

WBCHSE Applied AI · Work through your exam, one topic at a time.

All guide sections

Introduction to Artificial Intelligence

Domain 1 - Introduction to Artificial Intelligence. This domain tests whether you can define AI precisely, separate it from machine learning and deep learning, and classify real systems by capability and by technique rather than by marketing language.

Core rules and concepts

  • AI is the umbrella discipline: machines performing tasks associated with human intelligence - perception, reasoning, learning, language, decision-making.
  • Nesting: AI ⊃ Machine Learning ⊃ Deep Learning. ML learns parameters from data instead of following hand-written rules; DL is ML with multi-layer neural networks that learn features themselves.
  • Capability tiers: narrow AI (ANI) handles one bounded task; general AI (AGI) is a hypothetical human-level system that transfers across tasks; superintelligence is speculative. Every system in this syllabus is narrow.
  • Turing Test (1950): a behavioural test of indistinguishability in text conversation. It measures imitation, not understanding or consciousness.
  • Agent view: percept → internal representation → reasoning and planning → action. Classical AI emphasises search and knowledge representation; modern AI emphasises learning from data.
  • Milestones: Dartmouth workshop 1956 (the term AI is coined), expert systems in the 1980s, Deep Blue beats Kasparov at chess in 1997, AlexNet wins ImageNet in 2012, AlphaGo beats Lee Sedol in 2016, large language models from 2018 onward.

Worked example

A school dashboard flags students at risk of dropping out using attendance, marks and past outcomes. Is it AI? Yes - if the decision boundary is learned from historical labelled cases (dropped / not dropped), it is narrow AI, specifically supervised classification. If the dashboard only applies a fixed rule such as attendance below 60%, it is rule-based automation, not machine learning. The test to apply is always: were the parameters learned from data, or written by a human?

Traps that cost marks

  • Writing that AI means robots, or that AI means computers think like humans; the exam wants task-based definitions.
  • Treating AI, ML and DL as synonyms - ML is a subset, not a replacement.
  • Claiming the Turing Test proves consciousness or intelligence in general.
  • Calling Deep Blue a learning system: it won by search and evaluation functions, not by training on data.