Free AIF-C01 practice questions

AI, ML, and deep learning form a nested hierarchy from broadest to most specialized.

AI is any technique that makes a machine mimic human intelligence, ML is the subset that learns patterns from data, and deep learning is the subset of ML built on multi-layer neural networks, so option A orders them correctly. The other options reverse or break this nesting, which the exam treats as a clear error.

1. Start with the widest term: AI covers all machine intelligence, including rule-based systems.
2. Place ML inside AI as the data-learning subset, and deep learning inside ML as the neural-network subset.
3. Confirm the order runs AI, then ML, then deep learning from outside to inside.

Exam tip: AI contains ML, and ML contains deep learning.

A token is the small unit of text, often a word or part of a word, that a language model reads and produces.

Language models do not process letters or whole documents directly; they process sequences of tokens, so option A is correct. An embedding is a numerical representation of meaning, an access token is a security credential, and splitting a document is chunking, which is why the other options are wrong.

1. Recall that models work on sequences of small text units.
2. Match the definition of that unit to the term token.
3. Rule out embedding, credential, and chunking as different concepts.

Exam tip: A token is the small unit of text a model reads and generates.

RAG grounds a foundation model's answer in information retrieved from an external source at query time.

RAG searches a knowledge source for relevant passages and inserts them into the prompt as context, so the model answers from current or proprietary data instead of only its training. Option A captures this. The distractors describe fine-tuning (B), distillation (C), and an inference parameter (D), none of which retrieve external data.

1. Recall that RAG adds retrieved context to the prompt rather than changing the model.
2. Match the phrase 'answer from current or private data' to RAG.
3. Eliminate options that describe training or unrelated settings.

Exam tip: RAG retrieves relevant information and adds it to the prompt so the model answers from current or private data without retraining.

Veracity (truthfulness) is the responsible-AI feature about factual accuracy of outputs.

A model stating false information describes a truthfulness failure, which maps to veracity. Robustness (reliability under bad inputs), inclusivity (serving diverse people), and fairness (equitable treatment) are separate features that do not describe factual accuracy.

1. Identify the symptom: the model produces false statements presented as fact.
2. Match the symptom to the feature about accurate, factually grounded output.
3. That feature is veracity, also called truthfulness.

Exam tip: Veracity, or truthfulness, is the responsible-AI feature about outputs being factually accurate.

The shared responsibility model splits security into AWS's part (security of the cloud) and the customer's part (security in the cloud).

The customer secures their data, access, and configuration. Patching hardware, physical data center security, and the managed operating system all fall to AWS as security of the cloud, so those options are wrong.

1. Recall the split: AWS secures the cloud, you secure in the cloud.
2. Place each option on the correct side of the line.
3. Data, access, and configuration are the customer's half.

Exam tip: For AI workloads, you secure your data, access, and configuration while AWS secures the infrastructure.