Source
Adapted from my own handwritten and hand-typed notes from Year 1 to Year 3 of my EEE degree
Artificial intelligence is intelligence demonstrated by machines, in contrast to the natural intelligence of humans and animals. The field was born at the 1956 Dartmouth Conference, building on Alan Turing's work.
Four Views
| Human-like | Rational | |
|---|---|---|
| Thinking | Thinking humanly (cognitive modelling) | Thinking rationally (laws of thought) |
| Acting | Acting humanly (Turing test) | Acting rationally (rational agents) |
Thinking Humanly
Points:
- imitates human thought processes, tested by cognitive simulation
- came from the 1960s "cognitive revolution" in information processing psychology
- needs scientific theories of what happens inside the brain, now mostly the job of cognitive science and neuroscience
Acting Humanly: Turing Test
Turing (1950) proposed the imitation game: a human interrogator chats with a hidden human and a hidden AI. If the interrogator cannot tell which is the machine, the machine acts humanly. It suggested the main parts of AI: knowledge, reasoning, language understanding and learning.
Thinking Rationally
Points:
- follows the "laws of thought" from Aristotle's logic, a direct line through maths and philosophy to modern AI
- leads to rule based systems and logical inference
- problem: not all intelligent behaviour comes from logical deliberation
Acting Rationally
Rational behaviour means doing the right thing, the action expected to maximise goal achievement given the available information. This course is about designing rational agents.
State of the Art Milestones
| Year | Milestone |
|---|---|
| 1997 | Deep Blue beats Kasparov at chess |
| 2011 | IBM Watson wins Jeopardy |
| 2016 | AlphaGo beats Lee Sedol 4:1 at Go |
| 2017 | AlphaGo beats Ke Jie 3:0; Libratus beats top poker pros (game solving, not deep learning) |
Other areas: robot soccer, driverless cars, legged robots, voice assistants, manufacturing, finance. Open challenges are complex interactions: stochastic open environments, many players, sequential online decisions and strategic behaviour.
The rest of the AI part builds on Intelligent Agents.