Solving the AI Race Finalists — LessWrong