Innocuous data points can be combined to derive a revealing look at your life. Photo: Manu Fernandez / AP
The privacy debate tends to focus on how big companies handle "private" information like Social Security Numbers, credit histories, financial transactions and medical records—tangible info that can easily be used to get a peek at your life.
Intimate data: But according to University of Pennsylvania computer science professor Michael Kearns, the most valuable data is "intimate" data.
- That includes opinions, attitudes, beliefs and moods that aren't written down anywhere, but can be inferred from your online behavior — the posts you like on Facebook, the photos you share, the videos you watch, the items you buy on Amazon, your search queries, your location, etc.
- Advances in machine learning, deep learning and neural networks is making it easier to see patterns across raw data. That means otherwise innocuous data points can be combined to derive a revealing look at your life.
- "The most valuable data can't be measured in bits," Kearns said Tuesday at an AT&T-hosted privacy event in DC. Companies that have access to this information "can make all kinds of inferences about you and your life circumstances that you may not even know yourself."
Machine learning: Kearns says policymakers need to consult machine learning engineers, who have a detailed view of how data is being linked together. The FTC is interested in including those technologists in its policy discussions and is keeping an eye on developments in the data analytics and artificial intelligence areas, said Maneesh Mithal, associate director of the FTC's Division of Privacy and Identity Protection.
Our thought bubble: Even though a lot of tech and telecom companies say it's in their best interest to be transparent with customers about how their data is used, most people don't have a way to fully understand how their data is being pieced together and what these companies really know about them as a result.That's helping to drive a push for more awareness of data practices, even if privacy regulation — at least in the U.S. — is still pretty far off.