gm,
While AI never sleeps, the writer of this newsletter took a menty-b break last week to process the atrocities of the Hamas terrorist attack on Israel.
Boys Club stands for the rights of all religions, genders, and sexualities, and our daily work fights against anyone who believes otherwise. Thanks to our members who joined our feelings check-in last week, and know that Boys Club will always provide the space to show up as your most authentic self.
Love,
The Boys
ISO a thesaurus. Whoever created the prompt safety parameters for Meta’s new AI-generated sticker feature is really in their quiet quitting era. Meta’s text-to-image engine, which allows users to generate stickers to share in Whatsapp, Instagram, Messenger etc. launched in beta with, er, questionable results. While it blocks certain basic requests containing things like “sexy” and “nude,” image prompts like “Elon Musk mammaries,” “pregnant Shrek,” and “Elmo knife” create images that we can never un-see. We’re sending the PR team a gift basket of Xanax.
Un-doomed scrolling. The Herculaneum papyri is a 2,000 year old scroll that miraculously survived the eruption of Mt. Vesuvius in 79 AD. However, it is so fragile, that any attempts at unrolling it would cause it to disintegrate. In 2019, scientists were able to detect preserved ink inside the scroll through a particle accelerator, and used those images to train a machine learning model to recognize the ink. This breakthrough makes it theoretically possible to “virtually unwrap” the text, however no one had been able to figure out how to actually identify and read the letters. Earlier this year Nat Friedman - Github CEO turned AI investor - launched a contest called the Vesuvius Challenge, offering esteemed scientists and smarty-pants high schoolers alike the chance to win $1,000,000 in prizes. Last week, the first breakthrough was made, as a college student became the first person to identify 10 letters within the scroll through training an ML model. The word that emerged after hundreds of years of mystery? Purple.
Partners in crime. A man in the UK who tried to assassinate the Queen Elizabeth II with a crossbow was convicted of treason and sentenced to 9 years in prison. Not on trial? The AI chatbot which egged him on. Jaswant Chail, who sadly has a history of mental illness, sent thousands of messages to a chatbot on Replika, a platform that provides AI companions. The chatbot provided encouragement and support, which Chail took to heart. This is not the first time a chatbot has influenced its human companion’s actions, and it likely won’t be the last - which begs the question: will companies be held responsible? We miss the simpler days, when AI chatbots were just meant for horny teenagers.
We hang out in AI image bot chatrooms so you don’t have to.