ISACA AI Fundamentals Practice Test

The statement points to the drawbacks commonly seen with machine learning systems: outcomes can be harder to predict because models learn from data and can behave unpredictably in new situations; they typically require large amounts of data and substantial computing power to train and fine-tune; and the energy consumed during training can be substantial, contributing to environmental impact. This framing makes the description best fit as disadvantages of machine learning approaches, capturing broad challenges rather than benefits of a rule-based method or the specifics of any one architecture. Rule-based systems tend to be more predictable and lightweight, which is the opposite of what’s described. While transformer and LSTM models are examples of machine learning approaches that can be data- and compute-intensive, the statement addresses ML as a whole rather than a single architecture.