Operationalizing two tasks in Gary Marcus’s AGI challenge — LessWrong