NEW
Font size
WorksheetsOracle AI Foundations Quiz (Questions 1-15)
Total questions: 15
Worksheet time: 8mins
What is the primary difference between supervised and unsupervised learning?
Supervised learning uses labeled data; unsupervised learning does not.
Supervised learning is only for images; unsupervised is only for text.
Unsupervised learning always produces better accuracy.
Supervised learning requires no human input.
Which of the following is a key characteristic of Convolutional Neural Networks (CNNs)?
They use convolutional layers that preserve spatial relationships.
They cannot be used for image tasks.
They only work with sequential data like time series.
They have no trainable parameters.
Which model type is best suited for sequence tasks such as language or time series?
Feedforward neural networks
Sequence models (RNNs, LSTMs, Transformers)
K-means clustering
Decision trees
Generative AI systems are primarily used for which of the following tasks?
Only data cleaning
Creating new content (text, images, code) from learned patterns
Only model explainability
Enforcing data privacy
Which statement best describes a Large Language Model (LLM)?
A) A small model that performs arithmetic faster than humans.
B) A model trained on large corporal of text to generate or understand language.
C) A rule-based chatbot with no learned parameters.
D) A model that cannot be fine-tuned.
Which OCI capability is most relevant for deploying AI models at scale on Oracle Cloud?
OCI Block Storage only
OCI AI Services / OCI Data Science / OCI AI Infrastructure
OCI DNS
OCI Marketplace exclusively
What is ‘vector search’ typically used for in AI applications?
Searching plain text by exact match only
Finding semantically similar items using vector embeddings
Replacing relational databases entirely
Encrypting data for storage
Which of the following is a common metric for classification tasks?
Mean Absolute Error (MAE)
Accuracy, Precision, Recall, F1-score
BLEU score (only for images)
Throughput (network metric)
Prompt engineering for LLMs is primarily about:
Building physical prompts for robots
Crafting input text to steer model outputs toward desired responses
Optimizing GPU memory usage
Training the LLM from scratch
Which step is most important to reduce model bias and ethical risk before production?
Ignore bias — models are objective
Perform dataset audit, bias tests, and apply mitigation strategies
Increase model size only
Only deploy to test users and hope for the best
Which of these is a reason to use transfer learning?
To always avoid using pre-trained models
To fine-tune a model trained on large data for a specific, smaller task
To reduce accuracy on new tasks
To convert unsupervised problems to supervised automatically
A ‘skill check’ or lab in the OCI AI Foundations course typically helps students to:
Replace theoretical study entirely
Gain hands-on experience applying concepts on OCI resources
Avoid using cloud resources
Learn only marketing material
Deep learning models always require more data than simpler machine learning models to reach good performance.
True
False
Generative AI cannot be regulated because it’s impossible to trace outputs.
True
False
Embeddings convert items (text, images) into numerical vectors that capture semantic meaning.
True
False
