Historical customer attributes and churn or retention outcomes are available. Estimate churn risk for new customers as a class or probability.
A separate team wants product-description drafts from verified product facts, with human review before publication. Compare required outputs and evaluation goals rather than selecting only by a model name.
Choose TWO appropriate mappings to the required output.
Choose churn estimation and product-draft generation by required output and evaluation. Generative models can also classify, so do not treat predictive and generative use cases as universally exclusive model families.
Detailed explanation
Fluency does not establish factual accuracy. Preserve the scenario requirement for verified specifications and review before publication.
The goal is producing text under input constraints. Draft generation is separate from verifying facts and deciding whether to publish.
This estimates an unknown outcome from known examples. Provide data and evaluation aligned with churn risk.
Evaluation must match the requested output. Readability alone does not establish that churn predictions are useful.
LLMs can perform classification. Organize predictive and generative use cases by required output rather than assuming all model capabilities are mutually exclusive.
Check official documentation
Compare the following points with the official documentation.
Compare AWS supervised-learning and generative-AI descriptions with the Bedrock classification and text-generation examples. Identify required outputs, available data and evaluation, including classification by generative models and separate factual review of drafts.
Official references
- Official documentation 1
- Official documentation 2
- Official documentation 3
- Official documentation 4
- Official documentation 5
- Official documentation 6
What to confirm
Distinguish churn estimates from product-draft generation by required output. Do not categorically rule out classification by generative models or use fluency alone to guarantee prediction quality or factual product details.
Key points
- Required output
- Prediction and generation
- Classification capability
- Evaluation and review
Notes
- Environment: Official AIF-C01 Task 1.1 and AI/ML documentation review; no model training or production changes are performed.
- Compare the definitions with current AIF-C01 Task 1.1 and the linked primary sources.
- A conceptual example does not establish measured model accuracy or authorize external operations.
Foundation review
Start with the required output
Estimating a churn class or probability differs from drafting a product description. Compare business purpose, data and constraints rather than model names or popularity.
Avoid mutually exclusive capability assumptions
A generative model can also perform classification. A useful distinction between use cases is not a strict division applying to every model capability.
Match evaluation and pre-publication checks
Churn estimates need relevant predictive evaluation; product drafts need factual and wording checks. Fluent prose or one successful result does not guarantee correctness or business suitability.