WorksheetsGenerative AI & Prompt Engineering: Knowledge Check (SET-2)
Total questions: 10
Worksheet time: 3mins
An LLM fine-tuned specifically to follow commands and provide clear, concise answers tailored to the user's explicit request is known as a(n):
A) Base LLM
B) Instruction-based LLM
C) Probability Engine
D) Pattern Matcher
What LLM limitation is defined as the AI confidently generating fluent-sounding, but incorrect, fabrications?
A) Bias
B) Hallucination
C) No True Understanding
D) Static Data Limitation
The fact that LLMs are 'highly advanced pattern matchers' and do not 'think or understand in the human sense' leads to which critical challenge?
A) Hallucination
B) Bias
C) No True Understanding
D) Static Data Limitation
Why is outdated data a major concern for LLMs operating in high-stakes fields like finance?
A) It increases the model's parameters.
B) It prevents the model from generating complex code.
C) Access to real-time intelligence is critical for fraud detection and reacting to market shifts.
D) It makes the model less likely to suffer from bias.
Which key feature distinguishes the two core types of LLMs presented in the slides?
A) The number of parameters (billions vs. trillions).
B) Whether they predict the next word or follow commands.
C) The amount of data collected during pre-training.
D) The type of hardware they run on.
Which advanced tactic is crucial for integrating AI outputs into other applications or workflows by ensuring the response adheres to a specific format like JSON or a two-column table?
A) Few-Shot Prompting
B) Structured Output
C) Role Specification
D) Delimiters
The instruction to "internally plan or work out its solution first" for complex reasoning tasks is a key component of what core principle?
A) Structured Output
B) Give the Model Time to "Think"
C) Few-Shot Prompting
D) Use Delimiters
In the example: "Input: Happy -> Output: Joyful. Input: Tired -> Output: [Let Al complete]", what technique is being demonstrated?
A) Chain-of-Thought (CoT)
B) Zero-Shot Prompting
C) Few-Shot Prompting
D) Role Specification
The technique of forcing the AI to show its step-by-step logic before providing the final answer is called:
A) Few-Shot Prompting
B) Structured Output
C) Chain-of-Thought (CoT)
D) Fact-Check Mitigation
The master prompt example combines four advanced tactics. Which tactic ensures the prompt is delivered in a reusable table format?
A) Role Specification
B) Chain-of-Thought
C) Delimiters
D) Structured Output
