Bridging Expected Utility Maximization and Optimization — AI Alignment Forum