Is AI Gain-of-Function research a thing? — LessWrong