Exam-style
A homeware retailer approves four projects. (1) Marketing needs product shots for a catalogue nobody photographed, a fifteen-second clip for social, and a spoken voiceover. (2) Support wants every recorded call's transcript reduced to a paragraph a supervisor can skim. (3) Every inbound support email lands in one of eleven fixed queues, and email from a priority-contract account pages the duty manager instead. (4) The storefront wants an assistant a shopper can talk to, so "what do I need to start baking sourdough" comes back as a conversation naming catalogue products. Which one is not a job for a foundation model?
Reveal the answer
B. The third, because sorting email into eleven fixed categories is text classification for a purpose-built AI service, and the priority rule is an account-field lookup
The AI Practitioner exam guide lists the generative AI use cases as image, video, and audio generation; summarization; AI assistants; translation; code generation; customer service agents; search; and recommendation engines. Marketing’s work is image, video, and audio generation, and no single Bedrock model outputs all three: image-generation and video-generation models live on Amazon Bedrock, and Amazon Polly synthesises the voiceover. The call paragraph is summarization, the transcript supplied in the request. The storefront assistant is an AI assistant, a customer service agent, catalogue search in a shopper’s own words, and a recommendation engine. That leaves the third. One label out of eleven fixed queues is classification, and Amazon Comprehend trains a custom classifier on your labelled examples and reports accuracy, precision, recall and F1. The priority rule reads a tier off the account record and branches. Diffusion models are one family of foundation model, and the guide names recommendation engines.
Q. Catalogue images plus a clip and a voiceover, call transcripts reduced to a paragraph, inbound email sorted into eleven fixed queues, and a shopping assistant that talks. Which one is not foundation-model work?
A. The email routing. One label out of eleven is classification, not generation, and the priority-contract rule is a field lookup.
Why? Learn the list: image, video, and audio generation; summarization; AI assistants; translation; code generation; customer service agents; search; and recommendation engines. Three of these land on it. The catalogue work is image, video, and audio generation: an image model and a video model on Amazon Bedrock, with Amazon Polly for the voiceover. The call paragraph is summarization, safest because the transcript travels with the request. The storefront assistant touches four entries: AI assistant, customer service agent, catalogue search in a shopper’s words, and recommendation engine. Routing touches none. It produces a label, and Amazon Comprehend trains a custom classifier on your labelled examples and reports its accuracy and F1. The contract-tier branch is plain code. All four involve English text, so they feel alike. Sort by what comes out the far end instead. New content is a foundation model. One category out of a known set is a classifier. One row out of a table is a query.