If more than two responses are a good fit, or if no responses work, will randomize one.And the creators say this isn’t that unrealistic because, well, a person like this is usually half-listening to you anyway.Because of that, people are forced to come up with something generic to say in their first few exchanges, which leads to d.bot-like behavior.I and other women I enlisted tried out to see what exactly this fake dude was all about.Any time you send a message, an algorithm parses it for keywords and compares them to all of d.bot’s responses.The more your keywords match a response, the more likely that’s the response will use.
Referencing the Turing test, where a computer passes only if it can fool testers into thinking it's human, the creators took the approach of a bot that impersonates a very specific type of person.
Their creation has two purposes: One is to explore chatbots and artificial intelligence, and the second is to share a social message.
Hints of these experiences lie in d.bot’s microaggressions, including “Where are you from? Chin said submitting her own has been cathartic, and she's hoping that aspect of the project will grow. Instagrams like @byefelipe and the aforementioned @tindernightmares receive hundreds of submissions showing messages men have sent women.
In October, Mia Matsumiya entered the spotlight for her Instagram @perv_magnet, where she posts all the sexist and racist messages she’s saved over the course of a decade to call out online harassment. “I want my account to be a place where women can commiserate and men to just learn what women can experience online." The creators were partly interested in creating this chatbot to get at the “subtly chauvinistic or subtly prejudiced comments,” which Collinsworth said are less explored and just as important to acknowledge as the outrageous ones.
They're the kind that don't seem offensive on the surface, so d.bot’s messages range from innocuous to extreme.