The ongoing debate about exploiting information gathered from internet users highlights the importance of balancing the free flow of information and privacy protection. Finding a solution is not easy.

_______________________________________________________________________________________

By Jian Wang

(Wang, vice editor-in-chief of the New Media Division of China National Radio, is a visiting scholar at the Center for the Digital Future. Contact Wang at [email protected].)

Two internet companies were lashed hard by their respective regulators last April. In Washington D.C., Mark Zuckerberg was grilled in a two-day, 10-hour-long congressional hearing for Facebook’s privacy leakage incident. In Beijing, China’s widely used news site, Jinri Toutiao, was forced to shut down its most popular channel because of vulgar content and off-color jokes.

Seemingly the two companies didn’t commit the same sin. But underneath, artificial intelligence algorithms were the common factor that led to their awkward positions.

In Zuckerberg’s testimony, the Facebook CEO used “skiing” as an example to clarify Facebook’s business model, in which AI-driven algorithms are looming:

“If an advertiser comes to us and says, ‘Alright, I’m a ski shop and I want to sell skis to women,’ then we might have some sense because people shared skiing-related content or said they were interested in that,” Zuckerberg said.  “They shared whether they’re a woman. And then we can show the ads to the right people. ”

In another inquiry, Senator Bill Nelson also pointed fingers at AI-driven algorithms: “I’m communicating with my friends on Facebook, and indicate that I love a certain kind of chocolate. And, all of a sudden, I start receiving advertisements for chocolate.”

_________________________________________________________________________________________________

For internet companies, to “sense” user interests and find out “the right users” on the internet is like finding a needle in a haystack. Such a task done by humans alone would be not only impossible but also disastrous. In contrast, by employing the help of artificial intelligence, the internet companies surely secure efficiency. However, just like all other technologies that have changed society, be it atomic energy or genetic engineering, the devils unleashed by AI-driven algorithms are powerful.

_________________________________________________________________________________________________

For most users, their knowledge of AI algorithms is no more than what senators and congressmen demonstrated. However, the trepidation triggered is the same: How do advertisers know about me? What else do Facebook and advertisers know? Will my private information be used against me?

This worrying mood is also rampant for users of Jinri Toutiao. On Jinri Toutiao’s website, if you happen to click on an article about a movie star or view a picture of a sexy model, unstoppable celebrity gossip or vulgar sexual material may come your way.

Despite the widespread concern about privacy leakage, AI-driven algorithms are still considered one of the most promising technologies. In banking, education, retailing, transportation, and governmental services, AI-driven algorithms deliver incomparable efficiency. Users’ activities online are often collected as digital footprints.

This information, used on personal level, can make effective predictions on personal activities. When data is funneled into a big data pool, they can help streamline human activities such as production or transportation efficiently. It works just like Google Maps. Every vehicle using the map’s navigation is at the same time uploading data. Thus, the mega-data collected can make accurate predictions on city traffic flows and direct vehicles to less congested roads.

For internet companies, to “sense” user interests and find out “the right users” on the internet is like finding a needle in a haystack. Such a task done by humans alone would be not only impossible but also disastrous. In contrast, by employing the help of artificial intelligence, the internet companies surely secure efficiency. However, just like all other technologies that have changed society, be it atomic energy or genetic engineering, the devils unleashed by AI-driven algorithms are powerful.

First, the linkage of data sets destroys anonymity, which delivers protection as well as a sense of security to internet users. By linking age, gender, race, and other personal information, AI-driven algorithms often make anonymity void and meaningless.

Secondly, AI-driven algorithms not only can identify a user’s favorite food, commuting route, and purchasing habits, but also can make inferences about his/her political association, religious beliefs, sexual orientation, health condition, credit worthiness, and propensity to crime. It has been reported that some loan companies even tried to evaluate people’s credit worthiness based on the stores they frequently visit. Is that fair?

Thirdly, data’s persistence and the extent to which they are repurposed are often beyond our control. We typically have no way of knowing how long a company keeps our personal data and how our private information is used. What if our personal information is kept for a long time and used in ways with which we would not be comfortable?

Furthermore, we need to be cautious of data spillover. Living in the world, we are somebody’s spouse, child, parent, coworker, or friend. Sometimes we are just a pedestrian or a passerby. However, as another person’s data is being collected, our information may happen to be collected as well. For example, some apps that help remind a driver where his/her car is parked in a garage require the driver to take a picture of the vehicle before leaving. If your car is parked next to it, your license plate and personal location may be disclosed as well.

For a technology like AI-driven algorithms, we need to make best effort to reinforce its benefit and curtail the danger.

First, stringent consent requirements should be enforced to protect privacy. In this regard, the European Union is doing a much better job. A well-designed EU General Data Protection Regulation makes sure that users are aware of their rights to privacy and the power to say “no” whenever they feel insecure. Without users’ consent, no private information can be collected.

Second there should be a balance between privacy and convenience. The relevant procedures of internet companies should accurately reflect users’ willingness on the scale from “total protection” to “total convenience.”  Internet companies should not let AI-driven algorithms make assumptions about people’s willingness to receive ads. In regard to privacy, nothing should be taken for granted.

Finally, there should be an easy way for people to opt out. On apps, we often find it easy to register. But it is typically challenging to find a button to disconnect the service. Even if a user can delete an app, the private information already collected often cannot be deleted. The game should not be designed as “once on board, always on board.” Instead, there should be an exit whereby users can retrieve or delete all digital footprints.

Achieving a good balance between the free flow of information and privacy protection is always a struggle in this information age. Finding a solution is not easy. But before new rules can be put in place, traditional methods such as corporate ethics, legal regulations, stringent working procedures, and self-protection should be utilized. All of this will require collective effort from the internet companies, the users, and government. In this sense, Mark Zuckerberg is not alone in answering hard questions. All of us must work together to find adequate solutions.

__________

See all columns from the Center.

December 18, 2018