I posted my last post to the Effective Altruism Forum, where it received considerably more attention than I’d anticipated. On the one hand, this was nice, because I’d spent a decent amount of work on it and it felt good to have it appreciated. On the other hand, this was terrifying, because I really have no idea if my thesis was correct—I might’ve just convinced a bunch of people to adopt a worse communication strategy, thereby marginally decreasing the amount of resources going into AI alignment, thereby marginally increasing the likelihood of an AI apocalypse.