Part 4/13:
Harish underscores that responsible AI begins with collecting only what is necessary. Many organizations fall into the trap of data hoarding—gathering broad swaths of information simply because storage is cheaper. Instead, he advocates for a disciplined approach: collect only relevant data, and only when needed.
For instance, in a retail or financial setting, unnecessary collection of data such as detailed personal habits or extensive device metadata should be avoided unless critically relevant for the specific use case. Over-collection increases privacy risks and complicates compliance with regulations.