Our bet on whether the AI market will crash — LessWrong