A Bear Case: My Predictions Regarding AI Progress — LessWrong