The Mach-Zehnder interferometer is a famous experiment that demonstrates the wave-particle duality of light. It consists of a beam splitter that splits a beam of light into two paths, recombines them, and creates an interference pattern.
In the context of the Mach-Zehnder interferometer, the conclusion you mentioned is not entirely accurate. The experiment shows that the behavior of light depends on whether we have knowledge of which path the light takes or not, but it does not imply that light is always a wave or always a photon based on that knowledge.
When we don't have information about which path the light takes (i.e., it is in a state of superposition), it exhibits interference patterns characteristic of waves. This is because the different paths the light can take interfere constructively or destructively, creating a pattern of bright and dark fringes on a screen.
However, when we gain knowledge of which path the light takes, the interference pattern disappears. This is due to the introduction of "which-path" information, which disrupts the superposition and causes the wave behavior to collapse into a particle-like behavior. In this case, the light behaves more like individual photons, and the interference pattern is lost.
So, the conclusion of the Mach-Zehnder interferometer experiment is that the knowledge of which path the light takes affects the observed behavior, shifting between wave-like and particle-like characteristics. It highlights the wave-particle duality of light and demonstrates the sensitivity of quantum systems to measurement and observation.