by Ryan Kang ’25
Published Mar. 3rd, 2022
Sci-fi movies and stories often see artificial intelligence hunt or exterminate the human race. A recent experiment also uses AI in a potentially malicious way.
When given information on a person’s social tendencies online such as dates, times, duration, and type of digital communication, a computer was able to pick out and recognize an individual out of 40,000 others.
The AI identified 14.7% of anonymized individuals after a week of looking at their interactions. With an additional 20 weeks, the computer recognized 24.3% of individuals, suggesting that the accuracy of the computer increases with time. Furthermore, when the AI was also given the person’s contacts, the user was identified 52.4% of the time.
A different group of researchers found that a computer could connect articles to their author. When given 70,000 articles and 2,000 different authors, it was able to link the article to the author 27% of the time.
According to the General Data Protection Regulation, companies can sell one’s data as long as it is anonymized. However, given new AI technology, there have been rising concerns over the compromisation of an individual’s privacy.
There has also been mention of using AI technology to identify writers with harmful intentions to create order and lead to a better society.
Although the rate of being identified may seem low, the accuracy for prototypes of this technology are relatively high. Upon further development, the tracing technology will continue to be improved – and our level of anonymity will continue to decrease.