GCP Generative AI Leader Practice Question 84: Pretrained APIs Democratize AI
Google Cloud Generative AI Leader · Test 9 · question 4 of 10
A retail company has no machine learning engineers on staff but wants to add product photo labeling to its app this quarter, without collecting a training set or tuning a model. Which approach lets the team's existing developers add this capability with the least ML expertise required?
Show answer and explanation
Answer:
Pre-trained APIs are one of the ways Google Cloud democratizes AI: they package a ready-to-use model behind a simple API call, so a developer with no ML background can send an image and get labels back immediately, with no dataset to collect and no model to train or tune. Option C is the only path that avoids training entirely while still shipping the feature this quarter.
- A is wrong because training a custom model from scratch requires a labeled dataset and ML expertise the team does not have, and is far slower than calling an existing API. - B is wrong because fine-tuning still requires assembling a labeled dataset and ML skill to run and evaluate the tuning job, which contradicts the "least expertise, this quarter" constraint. - D is wrong because hiring a data science team to build a bespoke model is the most expensive and slowest option, and defeats the purpose of using a low-code, pre-trained capability.