As per databricks documentation, goal of incrementally and progressively improving the structure and quality of data as it flows through each layer of the architecture.

Most of the times, raw data is not useful and need to be cleaned or supplemented with other data set.

we can store it in one layer, but it’s easier to understand and manage if those are kept Separate. This can be done logical or physical. It’s really your choice. You are going to find use cases where someone might need to access bronze data for their gold use cases. There can be some data quality issue with it.