ISTQB CT-GenAI
GenAI in Test Analysis, Design and Implementation
Easy
Test analysis with GenAI
- Which input is typically suitable as a basis when an LLM is to help with test analysis?
- What is a typical work product of test analysis with GenAI?
- What does the accuracy of LLM results in test analysis depend on most?
- How can GenAI help with the test basis even before testing starts?
- An LLM breaks the story "As a member I want to extend my loan" into individual checkable statements. Which task is this?
- What information does an LLM need to order test conditions sensibly by risk?
- Why does an LLM map requirements to the existing test conditions?
- An input field accepts quantities from 1 to 50. Which test technique should an LLM suggest here?
- A prompt contains a user story as text and an interface sketch as an image. What is such a prompt called?
- Acceptance criteria are improved step by step using prompt chaining. What happens between two steps?
Test design and test implementation with GenAI
- What can an LLM suggest when generating test cases? (Several answers correct)
- Why use GenAI to create synthetic test data instead of using real production data?
- What should good synthetic test data cover? (Several answers correct)
- An LLM receives structured test cases and produces code for a test automation framework from them. Which task is this?
- New requirements are added. What can GenAI do with the existing test scripts?
- Which factors can an LLM take into account when scheduling test execution? (Several answers correct)
- Which keywords form the basic pattern of a Gherkin scenario?
- A few-shot prompt for Gherkin scenarios contains several examples of story, test conditions and scenarios up front. What do these examples show the model?
- After generating test cases, the model is asked to create a table mapping each acceptance criterion to the test cases. What is this for?
- A tester asks the model to suggest a better prompt for end-to-end test procedures. Which technique is he using?
Medium
Test analysis with GenAI
- A story reads: "Users can upload photos of a reasonable size." What would an LLM most likely criticize when reviewing the test basis?
- How can an LLM detect anomalies in requirements more reliably? (Several answers correct)
- From "Orders of 200 € or more are delivered without shipping costs", an LLM derives test conditions. Which one is most fitting?
- With equal likelihood: which test condition should an LLM prioritize highest?
- Which aspects can an LLM sensibly consider when prioritizing test conditions? (Several answers correct)
- An LLM reports that requirement R7 is not mapped to any test condition. What can happen if this goes unnoticed?
- An application passes through the statuses draft, submitted, approved and rejected. Which test technique should an LLM suggest?
- An upload accepts PDF, JPG and PNG; all other file types are rejected. Which technique should an LLM suggest?
- Why add an interface sketch as an image to a user story when an LLM is to generate acceptance criteria?
- Which additional details usefully complement a multimodal prompt for acceptance criteria? (Several answers correct)
- A prompt chain improves acceptance criteria in three steps. Which order makes sense?
- An intermediate result of the chain contains an obvious error. What is the best way to proceed?
Test design and test implementation with GenAI
- An LLM generates only positive cases for a registration form. What is the most targeted way to improve the prompt?
- A team needs test data for a health insurer's app. Which use of GenAI makes the most sense?
- For which kinds of testing can AI-generated synthetic test data be used? (Several answers correct)
- An LLM translates test steps into code for an automation framework. What matters most before checking it in?
- Test case B needs a customer created by test case A. What must an execution schedule created by an LLM take into account?
- A few-shot prompt produces Gherkin scenarios that do not match the test conditions. What is recommended?
- A workflow: generate test cases from acceptance criteria, check coverage via a table, then work out end-to-end procedures via a meta prompt. Which techniques are combined? (Several answers correct)
- An LLM is to prioritize a test suite while weighing risks and dependencies. Why must the result be checked particularly carefully?
Hard
Test analysis with GenAI
- Which inputs typically fit test analysis with GenAI? (Several answers correct)
- Story A: "Guests can order without an account." Story B: "Every order is assigned to a customer account." What should an LLM report when reviewing them?
- An LLM has prioritized test conditions without receiving any information on likelihood or impact. How should the result be assessed?
- For which project is an LLM-supported coverage analysis most worthwhile?
- A discount depends on three conditions: customer status, order value and voucher. Which technique should an LLM suggest?
- An LLM suggests a particular test technique for a story. What is the right way to handle this?
- A team compares acceptance criteria from several prompt variants: with and without image, with vague and with precise instruction. What should the comparison show?
- Which properties characterize well-formed acceptance criteria? (Several answers correct)
- Why refine acceptance criteria in several prompt steps instead of one single big prompt? (Several answers correct)
Test design and test implementation with GenAI
- Which task belongs to test design and not to test analysis?
- What is the core of test implementation?
- An LLM is to generate end-to-end test cases for a hotel booking: search, selection, payment, confirmation. Which addition to the prompt helps most?
- An LLM generates synthetic customer data. Which risks should be checked? (Several answers correct)
- A generated test script calls a method that does not exist in the framework being used. What is this most likely?
- What can GenAI generate from structured test cases? (Several answers correct)
- An LLM schedule puts a high-priority test at the start, although it needs data from another test. What follows from this?
- Which approaches can be compared against each other when prioritizing test cases with GenAI? (Several answers correct)
- What matters most when choosing the examples for a few-shot prompt for Gherkin scenarios?
- Which assignment is correct in a Gherkin scenario?
- Why is a meta prompt particularly worthwhile for end-to-end test procedures?