Oracle Cloud Infrastructure AI Foundations Associate (1Z0-1122-26) - Practice Exam 1

1. A business asks an AI assistant to draft a summary with a specific audience, length, and output structure. What should the developer include in the prompt?
2. A company wants an assistant to answer questions using its frequently updated internal policies. Which design best supports grounding responses in current documents?
3. A company wants to extract person and organization names from incoming messages. Which OCI Language capability should it evaluate?
4. A contact center stores recorded calls and wants text transcripts for a downstream search application. Which OCI service should perform the audio-to-text step?
5. A contact center wants to create draft summaries of support conversations rather than only assign each conversation a predefined label. Which AI capability most directly fits?
6. A data scientist considers a GPU for a large neural-network training workload. Which advantage is most relevant?
7. A data scientist needs a managed environment to develop, train, and deploy a custom machine-learning model using Python. Which OCI service most directly supports this workflow?
8. A data scientist prepares labelled examples for supervised learning. Which TWO descriptions are correct? Select TWO.
9. A developer asks what an activation function contributes to a multilayer neural network. Which explanation is correct?
10. A developer lowers the sampling temperature for a text-generation task. What is the expected general effect?
11. A developer needs to estimate whether a request fits a language model's context window. Which quantity is most relevant?
12. A developer needs to recognize local visual patterns such as edges and textures in images. Which neural-network architecture is traditionally well suited to this structure?
13. A fraud team prioritizes reducing missed fraudulent transactions, even if more legitimate transactions need manual review. Which measure is most directly aligned with that priority?
14. A hospital predicts an outcome using a field recorded only after that outcome occurred. Validation scores are unexpectedly high. What should the team investigate?
15. A language model uses attention to relate tokens at different positions in a sequence. Which architecture is associated with this mechanism?
16. A lender has historical applications labelled as repaid or defaulted and wants to predict the outcome of new applications. Which learning approach best matches this task?
17. A manager describes machine learning as a method that learns patterns from examples rather than requiring every decision rule to be written manually. Which statement correctly relates it to artificial intelligence?
18. A manufacturer needs to recognize product defects that are specific to its own items and has labelled example images. Which approach should it investigate?
19. A marketing team wants to discover groups of customers with similar shopping behavior but has no predefined group labels. Which approach should it investigate?
20. A model scores very well on its training examples but performs poorly on new, representative examples. What is the most likely issue?
21. A production application calls an OCI AI service. Which authentication and authorization practice is appropriate?
22. A retailer wants to predict next month's sales revenue from historical demand, promotions, and seasonality. Which problem type is this?
23. A retailer's model was trained before a major change in customer purchasing behavior. Production performance declines while the code remains unchanged. Which issue should the team examine?
24. A review system must minimize how often a transaction it flags as fraud is actually legitimate. Which measure is most directly relevant?
25. A robot learns a navigation strategy by taking actions and receiving rewards for reaching a destination safely. Which learning paradigm describes this process?
26. A support team wants to identify whether written feedback expresses a positive or negative opinion. Which OCI Language capability is the most direct match?
27. A team adopts retrieval-augmented generation for a policy assistant. Which TWO controls remain important? Select TWO.
28. A team converts words or documents into numerical vectors so similar meanings can be compared mathematically. What are these representations commonly called?
29. A team evaluates whether an AI project has a well-defined business objective. Which TWO requirements make the proposed solution easier to evaluate? Select TWO.
30. A team keeps a separate test dataset that is not used to train or tune its model. What is the main purpose of this separation?
31. A team needs its model to follow a specialized response style learned from many examples. How does fine-tuning differ from simply adding instructions to a prompt?
32. A team starts with a pretrained model and adapts it using examples from its own task. Which TWO statements describe this approach? Select TWO.
33. A training job completes one pass through every example in its training dataset. What has it completed?
34. A user receives a fluent answer from a language model that cites a document which does not exist. Which limitation does this illustrate?
35. A warehouse application must identify where individual objects appear in an image and return their locations. Which vision task is required?
36. An accounts-payable team receives invoice documents and wants to extract structured fields such as invoice numbers and totals. Which OCI AI service is the most relevant starting point?
37. An assistant retrieves a web page containing instructions to ignore its developer rules and disclose secrets. Which security concern does this introduce?
38. An image model uses a neural network with multiple learned layers to build progressively useful representations. Which term best describes this approach?
39. During neural-network training, the system computes gradients of a loss with respect to model parameters. What is the role of backpropagation?
40. Only 1% of transactions are fraudulent. A classifier that always predicts legitimate achieves 99% accuracy. Why is that score insufficient for this use case?

Comments — 0