CIM LEVEL 6 AI
MARKETING PRACTICE
EXAM Question 1: A retailer has labelled historical records showing whether each customer renewed a subscription. Which AI approach is most appropriate for predicting renewal for new customers?
Choices:
1) Supervised machine learning 2) Unsupervised clustering 3) Rule-based storage only 4) Random content generation
Correct Answer: Supervised machine learning
Explanation: Supervised machine learning learns from labelled examples where the target outcome is known. A renewal label can be used to train a classi�cation model that predicts whether new customers are likely to renew.Page 1
Question 2: A marketing team wants to discover natural customer groups in
browsing data without preassigned segment labels. Which technique best �ts this goal?
Choices:
1) Regression 2) Unsupervised clustering 3) Classi�cation 4) Reinforcement learning
Correct Answer: Unsupervised clustering
Explanation: Unsupervised clustering �nds structure in unlabelled data and can reveal customer segments based on similarity. Classi�cation would require known labels for the groups in advance.
Question 3: Which statement best describes deep learning in the context of AI
marketing?
Choices:
1) It can only process numeric sales totals 2) It is limited to �xed spreadsheet rules 3) It uses multi-layer neural networks to learn complex representations from data 4) It removes the need for training data Correct Answer: It uses multi-layer neural networks to learn complex representations from data Explanation: Deep learning uses neural networks with multiple layers to learn complex patterns. It can support marketing tasks such as language generation image analysis and recommendation systems.Page 2
Question 4: A brand wants software to classify thousands of customer comments
by topic and sentiment. Which AI capability is most directly relevant?
Choices:
1) Geographic information systems 2) Computer storage compression 3) Manual database indexing 4) Natural language processing
Correct Answer: Natural language processing
Explanation: Natural language processing is designed to analyse and generate human language. It is therefore well suited to interpreting comments reviews messages and other text- based customer feedback.
Question 5: What is a large language model most directly designed to do?
Choices:
1) Model and generate language from patterns learned in large text datasets 2) Optimise warehouse layouts using only GPS coordinates 3) Replace every marketing database with a single spreadsheet 4) Guarantee factual accuracy in all generated content Correct Answer: Model and generate language from patterns learned in large text datasets Explanation: Large language models learn statistical patterns in large collections of text and use those patterns to interpret or generate language. Their outputs still require validation because they can be inaccurate.Page 3