Question 32
Section 1Which statement best defines a foundation model in the context of generative AI?
Correct answer: A
Explanation
A foundation model is a "large model pre-trained on broad data" and then adapted for different uses. Its defining feature is generality: it can be fine-tuned or prompted for "many downstream tasks" rather than built for only one task.
Why each option is right or wrong
A. A large model pre-trained on broad data that can be adapted to many downstream tasks
Under the standard generative-AI usage adopted in major frameworks such as the Stanford CRFM definition and the NIST AI RMF terminology, a foundation model is characterized by scale, broad pre-training, and transferability to multiple tasks. The defining features are that it is trained first on large, diverse datasets and then reused via prompting or fine-tuning for different downstream applications, rather than being limited to a single narrow function.
B. A small task-specific model trained on labeled data for one classification problem
Describes a narrow supervised model for one task, not a general reusable pre-trained model.
C. A rule-based system that uses if-then logic and curated knowledge bases
Refers to symbolic AI with hand-written rules, not learned generative model behavior from large-scale pretraining.
D. A federated learning model that trains only on edge devices
Defines a training setup across devices, not the meaning of a foundation model itself.