When traces are left of an individual’s mobile phone interactions, as well as those of their contacts, artificial intelligence can track the person among more than 40,000 anonymous users. Researchers from various branches of the human sciences such as sociology point out that humans socialize in ways that allow algorithms to identify them by behavioral patterns even if they remove all their traces when browsing the Internet.
It is not surprising that people tend to stay within established social circles and that these regular interactions form a stable pattern over time, but the fact that such a pattern can be used to identify the individual is a twist of the nut. science fiction movies.
Everything stays on the net
Under the European Union’s General Data Protection Regulation, companies that collect information about people’s daily interactions can share or sell this data without users’ consent. The problem is that the data must be anonymized. Some organizations may assume that they comply with this standard by giving users pseudonyms, however in most cases this is not true.
There have been experiments hypothesizing that people’s social behavior could be used to track data sets containing information about their online interactions . To test their hypothesis, the researchers taught an artificial neural network, an artificial intelligence that simulates the neural circuitry of a biological brain, to recognize patterns in users’ weekly social interactions.
For a test, the researchers trained the neural network on data from a mobile phone service detailing the interactions of 43,000 people over 14 weeks. This data included the date, time, duration, type (call or text) of each interaction, the pseudonyms of the parties involved, and who initiated the communication.
The interaction data of each user was organized in web-like data structures consisting of nodes representing the user and their contacts. Threaded chains with interaction data connected the nodes. The artificial intelligence was shown the interaction network of a known person and then launched to search the anonymous data for the web that most resembled the one that the user had consumed.
The neural network only matched 14.7% of people with their anonymous selves. However, artificial intelligence identified 52.4% of people when given not only information about the target’s interactions, but also those of their contacts. When the researchers fed the algorithm target and contact interaction data collected 20 weeks later, the AI correctly identified users nearly 70% of the time, suggesting that social behavior remains identifiable over long periods of time. weather.
REFERENCES :
Intersoft Consulting/University of Minnesota
