Dilemmas
- AI progress originating from data growth or high quality dataset simply cannot exempt us from violating privacy;
-
We need relative and representative data in all research so that it is applicable in real-life;
- Not just AI research that is data-hunger because all research collaborates with each other at some point
Information Services
A typical narrative about privacy:
- Lack of data sharing could lead to service lock-in;
- More privacy violation could happen because of data-sharing;
But there is a difference regarding private data being transferred:
-
Private data moving with the owner;
- This is the sharing we want
-
Private data being sold with owner’s consent;
- This is not the sharing we want
We need Interoperability (e.g. install third-party apps on your phone, bypassing ink cartridge in printers) to fight against lock-in. In this case, we need this type of information flow that are in our control.