What is artificial intelligence
Definition
AI is the branch of computer science chasing problems we associate with human intelligence: image creation and recognition, speech to text, learning, reasoning.
The basic loop: a data scientist feeds a training dataset to an algorithm, and out comes a model. Users then throw new input at the model (“what is this?”) and get predictions back (apple, banana, peach).
The nesting to memorize
AI ⊃ Machine Learning ⊃ Deep Learning ⊃ Generative AI. What most people picture when they say “AI” (ChatGPT, DALL-E) is the innermost ring.
History in six beats
- 1950s: the Turing Test is proposed; McCarthy coins “artificial intelligence.”
- 1970s: expert systems (MYCIN diagnosing bacterial infections from rules).
- 1990s: machine learning and data mining take off.
- 1997: Deep Blue beats Kasparov at chess.
- 2010s: the deep learning revolution; AlphaGo beats Lee Sedol (2016).
- 2020s: AI everywhere: assistants, autonomous vehicles, diagnostics, plus the ethics debates.
Use cases
Transcription and translation, game playing, self-driving vehicles, speech recognition and generation, code suggestions, medical diagnosis, business process automation, fraud detection.
The worked example the course keeps returning to: Intelligent Document Processing. A scanned invoice PDF goes in, AI (computer vision + deep learning + NLP) extracts the seller, buyer, and line items, and the structured result lands in a database.