What does "garbage in$,$ garbage out" refer to in terms of algorithms in data processing and computation?

Algorithmic predictions cannot be accurate

Human input is unnecessary in algorithms

The quality of input training data influences the output

Algorithms are inherently biased

All of the choices