AIF-C01 mock exam

Question 1, Fundamentals of AI and MLA model scores 99% on training data and 70% on new data. What is happening?

• AUnderfitting
• BOverfitting
• CData drift
• DTokenization error

Question 2, Fundamentals of AI and MLYou have customer records with no labels and want to find natural groups. Which approach fits?

• AReinforcement learning
• BSupervised classification
• CRegression
• DUnsupervised clustering

Question 3, Fundamentals of GenAIWhat does a higher temperature do to a foundation model's output?

• AMakes it cheaper per token
• BMakes it longer on average
• CMakes it more varied
• DMakes it faster to generate

Question 4, Fundamentals of GenAIWhy are text embeddings useful for search?

• AThey encrypt documents at rest using a model key
• BThey translate text between languages before search
• CThey place similar meanings near each other as numbers
• DThey compress files so storage costs less overall

Question 5, Applications of Foundation ModelsA company wants a model to answer from its own frequently changing policies. Which approach is the most practical?

• AFine-tune the model every time a policy changes
• BUse RAG to retrieve current policies at request time
• CWrite every policy into the system prompt
• DTrain a new model from scratch

Question 6, Applications of Foundation ModelsWhich AWS service gives you API access to foundation models without managing infrastructure?

• AAmazon Comprehend
• BAmazon Bedrock
• CAmazon Kendra
• DAmazon Rekognition

Question 7, Guidelines for Responsible AIA team must block a chatbot from discussing competitors and remove personal data from replies. What should they use?

• AAmazon Bedrock Guardrails
• BA bigger foundation model
• CAmazon S3 versioning
• DAWS Cost Explorer

Question 8, Security, Compliance, and Governance for AI SolutionsUnder the shared responsibility model, who is responsible for what your application sends to a foundation model?

• AThe model provider
• BNobody, as it is transient
• CYou, the customer
• DAWS