Philosophies & Movements
Alignment
The problem of getting an AI system to actually do what its makers and users want — and not do things they don't want. 'Aligned' AI follows human intentions and values; 'misaligned' AI pursues goals that conflict with them, whether subtly or dangerously.
Origin · no single documented coiner
In AI safety the term was popularized by Stuart Russell and by Eliezer Yudkowsky's writing on the Machine Intelligence Research Institute's 'aligned AI' research program in the 2010s. The word itself is older engineering usage for matching a system to a specification.
No single primary source link for this term.