Basics of Generative AI and LLMs CT-GENAI Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   What is a core principle of how generative AI learns to create new content?

   • AIt relies on a fixed set of rules defined by human experts to generate content.
   • BIt requires labeled data for every possible output to generate correct content.
   • CIt memorizes all training examples and reproduces them exactly when prompted.
   • DIt learns the underlying patterns and structure of training data and uses that to produce new, similar content.
   Reveal Solution Discussion Select an answer first
2. 12expert · hard

   A QA engineer is using an LLM to generate a summary of a long test report. The engineer notices that the summary omits a critical failure. What is the most likely reason for this omission?

   • AThe LLM is not capable of summarizing long documents
   • BThe LLM's tokenization process removed the failure from the text
   • CThe LLM's context window may not have included the entire report, so it missed the critical failure
   • DThe LLM is biased against reporting failures
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   Which of the following is a common limitation of LLMs?

   • AThey can reflect or amplify biases present in their training data.
   • BThey require a human to verify every single output they produce.
   • CThey always provide the same output for the same input.
   • DThey are unable to process text longer than a single sentence.
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   Which statement best defines generative AI in the context of software testing?

   • AAI that predicts the probability of a defect based on historical test metrics.
   • BAI that classifies existing test results into pass or fail categories.
   • CAI that optimizes test execution schedules to reduce overall runtime.
   • DAI that creates new content, such as test data or test cases, based on learned patterns from existing data.
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A QA engineer is using an LLM to summarize a long bug report. The model processes the text by breaking it into smaller units. What is the correct term for this process and what does the model do with these units?

   • AParsing; the model builds a syntactic tree to understand the grammatical structure of the text
   • BEmbedding; the model converts tokens into high-dimensional vectors and averages them to produce a summary
   • CNormalization; the model removes stop words and punctuation to focus on key terms
   • DTokenization; the model converts text into tokens and uses them to predict the next token in the sequence
   Reveal Solution Discussion Select an answer first