The defining conceptual shift from traditional programming to Machine Learning is the source of the logic. In traditional programming a human writes explicit rules (IF/ELSE statements, lookup tables, formulas) and the computer applies those rules to input data to produce output. In Machine Learning the human provides input data plus desired outputs, and the algorithm learns the rules. The output of training is a model — a parameterised function that maps new inputs to predictions.

This is why ML is the right tool when the rules are too numerous, too nuanced, or too adaptive for a human to codify by hand — exactly the fraud-rules situation in the stem. AIF-C01 task 1.1 tests this recognition repeatedly. Option A is the only answer that captures the paradigm shift; the other options mix in implementation details (hardware, encryption, dataset size) that are not the conceptual boundary.

Plain-Language Explanation:

Traditional programming is like giving a new cook a laminated recipe card with exact steps. Machine Learning is like letting a cook taste thousands of dishes over a career until their palate internalises what 'balanced' means. The laminated card is written by a human. The palate is learned from examples. The shift from card to palate is the shift from rule-based code to ML.

Option Analysis:1309

A：Correct because the learned-from-data versus authored-by-humans distinction is the defining line between the two paradigms.
B：Wrong because ML usually benefits from more data, not less, and dataset size is not the conceptual boundary.
C：Wrong because ML algorithms run on both CPUs and GPUs; hardware choice is an implementation detail.
D：Wrong because encryption is unrelated to whether the system learns from data or applies hand-written rules.