What People Mean When They Say 'AI Agent'

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-builders-foundations-ai-agentic-workflows-explained-r9a10-teen

1. What two main components must be combined to create an AI agent?

   1. A chatbot interface and a database of answers
   2. A code editor and a version control system
   3. A large language model and a set of tools the model can call upon
   4. A search engine and a content filter

2. Which of the following is mentioned in the lesson as a general-purpose web agent?

   1. Claude Code
   2. Manus
   3. Devin
   4. OpenAI's Operator

3. Why should AI agents be given write access (like sending emails or making purchases) slowly and carefully?

   1. Because agents cannot learn to use write tools properly
   2. Because read-only tools are more advanced than write tools
   3. Because agents might make expensive mistakes like spending money or submitting the wrong form
   4. Because write access makes the agent run slower

4. Which capability is unique to agents like OpenAI's Operator compared to basic chatbots?

   1. Answering general knowledge questions
   2. Generating creative stories and poems
   3. Using a real web browser to book restaurants, order groceries, or fill out forms
   4. Translating text between languages

5. What is the minimum number of distinct components needed to build a simple AI agent, according to the simplest example in the lesson?

   1. At least five: an LLM, a search tool, a math tool, a loop, and an API
   2. At least three: a chatbot, a database, and an interface
   3. At least four: thinking, deciding, calling, and reading
   4. At least two: an LLM and one tool

6. What does the lesson mean when it says agents can 'autonomously' complete tasks?

   1. The AI can make decisions and take actions on its own without asking for permission for each step
   2. The AI is fully self-aware and understands its own reasoning
   3. The AI can learn completely new skills without any training data
   4. The AI can function without any internet connection

7. A coding agent like Claude Code would typically perform which sequence of actions when fixing a bug?

   1. Copy existing code from one file to another
   2. Generate a description of what the code should do
   3. Read the codebase, write changes, run tests, and iterate based on results
   4. Ask a human developer to write all the code

8. What makes a task an 'agentic' task rather than a standard AI generation task?

   1. The AI uses a more expensive language model
   2. The AI must use external tools and take actions in multiple steps to complete the goal
   3. The AI generates text that is longer than usual
   4. The AI processes images instead of text

9. What type of tool would be the safest to give an AI agent access to first?

   1. Read-only tools like search engines and file viewers
   2. Tools that can delete files
   3. Tools that can make payments
   4. Tools that can send emails

10. What happens when an AI agent 'reads the result' after calling a tool?

    1. The agent opens and reads a text file from the computer
    2. The agent learns new information for its next training iteration
    3. The agent processes the output from the tool (like search results or calculation answers) to inform its next decision
    4. The agent displays the result to the user

11. Why might using an AI agent to book travel or order food be considered risky?

    1. The agent might refuse to work with certain websites
    2. The agent might generate inappropriate content
    3. The agent might become too expensive to run
    4. The agent might select the wrong restaurant, enter incorrect details, or spend money unexpectedly

12. The lesson describes which simple prompt as already being an agentic loop?

    1. Who wrote Romeo and Juliet?
    2. Tell me a joke about computers
    3. What is 2+2?
    4. Use the web to research X, then write me a 3-paragraph summary with sources

13. What does the lesson identify as the core difference between an 'agent' and a typical chatbot?

    1. Agents are only available as paid services
    2. Agents cannot make mistakes
    3. Agents can take actions using tools, while typical chatbots only generate text responses
    4. Agents are always more accurate than chatbots

14. What is the relationship between an LLM and tools in an AI agent architecture?

    1. The LLM is unnecessary if you have enough tools
    2. The tools generate the text and the LLM decides which tool to call
    3. The LLM acts as the 'brain' that decides when and how to use tools, while tools extend what the LLM can do
    4. The LLM and tools operate independently without communicating

15. Based on the lesson, what is the current state of AI agent autonomy?

    1. AI agents have achieved full human-level autonomy
    2. Real autonomy is still limited, though the basic agent loop runs on millions of computers
    3. AI agents no longer need any human supervision
    4. AI agents can only work in very limited research settings