Quizzes

Getting Started quiz

8 questions from the Getting Started track. Score 80% or higher to unlock a certificate.

1. Welcome to Joeven

   1. What is Joeven primarily teaching?

   • Website CSS tricks and page layout
   • Autonomous AI agents, from first principles to production
   • Only prompt-engineering jokes and viral screenshots
   • How to install GPU drivers

2. What Is an AI Agent?

   2. Which statement is most accurate?

   • The language model by itself is the agent
   • The agent is the loop: goal, observations, actions, and a policy (often an LLM)
   • Agents are forbidden from using tools
   • Autonomy means you skip tests

3. The Agent Loop

   3. What should stop an agent loop?

   • Nothing — let it think until the process dies
   • A success check, a max-step or max-cost limit, or a human handoff
   • Only a keyboard interrupt from whoever is watching
   • Deleting the system prompt so the model forgets the goal

4. Agents vs Chatbots vs Workflows

   4. You must extract fields from PDFs, then write rows to a database. What should you build first?

   • A multi-agent swarm that debates each PDF
   • A workflow with an LLM extract step and schema validation
   • A chatbot with no tools and no database write
   • An infinite ReAct loop until the model feels done

5. The Learning Path

   5. Why does Joeven teach math before RAG?

   • To make the syllabus look longer
   • Because retrieval quality is geometry and probability, not just API calls
   • Math is required by the domain registrar
   • Transformers cannot run unless you finish calculus homework

6. Setup: Python, Keys, and Safety

   6. Where should production API keys live?

   • Inside the system prompt so the model can “remember” them
   • In a public GitHub repo so teammates can copy them
   • In environment variables or a secret manager, never in git
   • In the left sidebar of Joeven

7. Tokens, Tools, and Goals

   7. Why do agent loops get expensive quickly?

   • GPUs dislike the Python language
   • Each step resends growing context, so token use scales with steps
   • Tools are billed per millisecond of silent thought
   • JSON is not allowed inside prompts

8. Your First Tiny Agent

   8. In the tiny agent, what is the policy?

   • The JOBS dictionary with statuses
   • The max_steps budget integer
   • fake_model (later a real LLM) choosing the next tool
   • The print statements in the loop