A team of researchers at the Massachusetts Institute of Technology in Boston created what they've dubbed the "world's first psychopath AI."
The artificial intelligence algorithm, nicknamed "Norman" after the character Norman Bates in Alfred Hitchcock's Psycho, generates captions for Rorschach-style inkblot psychological tests. Because the researchers trained the algorithm on violent content from what they call "the darkest corners of Reddit," Norman discerned more disturbing descriptions than a standard image captioning model.
In one example, a standard image captioning model described an inkblot as "a group of birds sitting on top of a tree branch." Norman provided the caption "a man is electrocuted and catches to death."
In the wake of media reports on Norman, many people took to Twitter to express their surprise at MIT's creation, according to Time. However, the researchers stressed their reason for developing Norman was to underscore a foundational principle of AI — data matters.
"Norman is born from the fact that the data that is used to teach a machine learning algorithm can significantly influence its behavior," the researchers wrote on Norman's website. "So when people talk about AI algorithms being biased and unfair, the culprit is often not the algorithm itself, but the biased data that was fed to it.
"The same method can see very different things in an image, even sick things, if trained on the wrong (or, the right!) data set. Norman suffered from extended exposure to the darkest corners of Reddit, and represents a case study on the dangers of artificial intelligence gone wrong when biased data is used in machine learning algorithms."
Although the project focuses on the importance of data, the researchers chose to redact the name of the subreddit they used to train Norman, citing its "graphic content." They specified the subreddit's users submit written captions describing graphic images and videos about death, CNN reports.
To view Norman's inkblot captions, click here.