Jonas Hallgren comments on Universal AI Maximizes Variational Empowerment: New Insights into AGI Safety