Question 1

You want to reduce hallucinations in a policy‑QA bot. What is the FIRST step?

Accepted Answer

Ground answers in an internal knowledge base via RAG

Answer

Disable the model’s internet access

Answer

Use only small models

Answer

Force very short responses

Question 2

Your LLM sometimes gives correct answers but in random formats. What is the BEST fix?

Accepted Answer

Add examples of the exact JSON or table format you want (few‑shot)

Answer

Increase temperature

Answer

Call a larger model with the same prompt

Answer

Use more exclamation marks in the prompt

Question 3

As agentic AI scales, what is the human’s MOST realistic role?

Accepted Answer

Editor‑in‑chief/orchestrator: define tasks, guardrails, and review high‑impact actions

Answer

Only write complex code

Answer

Just monitor, AI runs everything

Answer

Manually review every AI response

Question 4

Meta just acquired Moltbook on March 11, 2026. What is Moltbook?

Accepted Answer

A social network designed for AI agents to interact

Answer

A code editor for LLM apps

Answer

A note‑taking app for prompt engineers

Answer

A GPU manufacturer

Question 5

According to recent reports, what is the big shift in this new wave of models (Gemini 3.x Pro, GPT‑5.x, DeepSeek V‑series, etc.)?

Accepted Answer

They move from “chatty” models to much stronger reasoning, code, and problem‑solving performance

Answer

They only run on mobile phones

Answer

They are mainly focused on offline use

Answer

They no longer support free access mode

Question 6

Thousands of elderly people recently lined up in China so that Tencent engineers could help them install it. What is the name of this exciting open-source AI agent?

Accepted Answer

OpenClaw

Answer

Moltbook

Answer

NanoClaw

Answer

Claude Skills

Question 7

What role is MCP increasingly playing in 2025–2026 agent ecosystems?

Accepted Answer

A standard “plumbing layer” that lets agents connect to tools, databases, and APIs in a uniform way

Answer

A new GPU chip from NVIDIA

Answer

A programming language for front‑end UI

Answer

A social media app

Question 8

In our Hallucination Taxonomy, what type is it when the model invents a citation that doesn't exist?

Accepted Answer

Fabrication

Answer

Grounding Failure

Answer

Factual Error

Answer

Context Drift

Question 9

An LLM returns: {"tool": "weather", "city": "Mumbai"} to call a tool. This is an example of:

Accepted Answer

Function/Tool Calling via structured output

Answer

RAG retrieval

Answer

Chain-of-Thought reasoning

Answer

Prompt injection

Question 10

What is Perplexity Computer best described as?

Accepted Answer

An “always‑on” digital worker that orchestrates many models and tools to run end‑to‑end workflows

Answer

A browser plugin only for search

Answer

A new standalone LLM model

Answer

A GPU driver

Related quizzes

Related quizzes