Hi Brave Team,
I am interested in your opt-in advertising idea, which looks like an interesting pro-privacy manner for delivering ads. This question relates to your June announcement of the concept, where it says, “It sends a detailed log of the browsing activity to Brave, which is used as algorithmic test data to check our on-device machine learning.” I know that ML models need large amounts of data to be trained to deliver ads, so I’m curious how the system will is trained initially and how it can continue to be trained to serve ads more effectively.
AFAIK, targeted advertising relies on creating individual user profiles based on past browsing data. Is there a way to do this such that your profile will remain your own and not passed around by centralized advertising companies, while still being useful for delivering targeted ads? Is it possible for ML models to improve without centralized data collection? I don’t claim any expertise in this topic, just genuinely curious as to more of the technical details for how such a system could work.