This paper discusses the application of artificial intelligence (AI) and machine learning (ML) in portfolio optimization and compares them to their performance against other such models as the mean-variance model that Markowitz developed. Though classical models suppose that financial markets are normal, linear, and exhibit fixed risk-reward trade-offs, the real world financial market structures are nonlinear, high-dimensional, and interdependent as a result of complex interactions that are hard to meet the assumptions. The current tools that can be applied to manage such complexities and adapt to the rapidly changing market environment are AI and ML methods, which are neural networks, reinforcement learning, support vector machines, and ensemble algorithms. The research study compares the efficiency of artificial intelligence-based tools to maximize risk-adjusted returns, diversify more, and respond to the negative risks with the traditional optimization models. As practiced empirically, the ML-based models are more efficient than the traditional ones in volatile and data-sensitive environments, notably when combined with other sources of data, such as the sentiment indices and the macro-economic indicators. The results show the possibility of the applications of AI-enhanced portfolio strategies to provide higher returns on investments and advance SDG 8 (Decent Work and Economic Growth) as it will help to allocate capital optimally and SDG 9 (Industry, Innovation, and Infrastructure) as it will introduce technology innovation in financial systems. This paper also concludes that AI and ML is not simply a white and refine the already existing state of portfolio optimization but a paradigm shift and that radical frontiers are in sustainable efficient financial markets.