Social networking services can use AI algorithms to analyze photos, and combine results with other information that they have collected over time, to extract much more information than what we want to give them. Further, such information might be shared with advertisers and other parties. I will introduce this problem, and have a brief discussion with participants to get a broader view of this problem.
I will then describe two possible solutions that I have been working on the past year. Adding artifacts or effects to photos can degrade the ability of AI algorithms to detect and recognize content in photos. AI algorithms built into smartphone apps can remove information that we do not want to share. These will be explained with a non-technical audience in mind. I will conclude the session with a brief discussion on how to proceed with these approaches for better privacy.
Some of the solutions that I plan to demonstrate in this session need further improvement to be accepted by ordinary users. My main intention in sharing my ideas and work in this session is to form a group of individuals who share the similar concerns. This will allow us to create a better understanding of the problem, and continue this work as a group.
How will you deal with varying numbers of participants in your session?:With a larger audience, I would keep the introduction and descriptions of work unchanged, but the discussion time can be shortened with a Slack channel created to continue discussions. With a smaller audience, I could make it more interactive, take examples and suggestions as well.
I am a former university academic who now works as a researcher in Computer Vision and Artificial Intelligence. I also participate in volunteer activities when I have time.