Artificial Intelligence
The field of making machines do things that seem to require intelligence — a definition that has moved every time the machines succeed.
When not to use it
- As a technical description. "We use AI" tells a listener nothing about what the system does or how it fails.
- As a reason to build something. The question is what problem you're solving, not which technique is fashionable.
- As a claim of understanding. Capability at a task is evidence about the task, not about comprehension.
Reach for something else instead
- Say what it actually is — a classifier, a language model, a rules engine. Precision costs nothing and prevents a lot.
- Traditional software when the rules are known. If you can write the condition, write the condition.
- Statistics when you want to understand a relationship rather than predict a value.
Sources & further reading
- Turing (1950), Computing Machinery and Intelligence — the question, and the test, from the beginning.
- Russell & Norvig, Artificial Intelligence: A Modern Approach — the standard text, and the clearest account of what the field contains beyond ML.
- McCarthy et al. (1955), A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence — where the term was coined, and worth reading for how confident it was.
Primary sources, listed so you can check the claims on this page rather than take them on trust.
Where people go wrong
- Treating AI and machine learning as synonyms. ML is a subset; plenty of working AI was never trained on anything.
- Assuming capability implies generality. Passing a benchmark demonstrates performance on the benchmark.
- Letting the word carry the argument. "It's AI" is a description of a technique, not evidence it works.
At a glance
FieldFoundations
Core ideamachines doing what seems to need intelligence
Definitionmoves whenever the machines succeed
Containsmuch more than machine learning
DifficultyBeginner
Flashcards for this concept
Question
—
Answer
—
1 / 4
Often compared with
AI vs. machine learning — the goal vs. the technique that currently dominates it.