Can the biggest problems in AI be solved by philosophy?