Why Liskov Substitution is needed

I am trying to gain more traction on Liskov Principle and every example I find seems to be screaming about how developers should write the program in a way that if the consumer of the software ends up using the derived class there shouldn't be a change in the behavior.

What does this substitution help? Is my understanding correct in a sense that LSP is required to add extensibility to an application or there is more to it?

If something like this has previously been answered, please redirect me and I will close the question.