Module 1 · 10 questions · Retry allowed

Module 1 Knowledge Check

Choose the best answer from the facts in each question. The self-check standard is 8 out of 10.

Browser-only self-checkThis page calculates your result in the browser. AabhaAI does not receive or store your answers, and the result is not a certificate or formal credential.

1. Which fact is the best proof of machine learning? The product speaks in natural language. A model was trained from examples and tested on new examples. The result appears automatically. The product name contains AI.

2. What is deep learning? One type of machine learning that uses neural networks with many layers. Every kind of automation. Another name for search. Only AI that creates text.

3. A service finds policy passages, writes a draft with a language model, and requires officer approval. What is the best description? Search, writing, and approval are the same thing. It cannot use AI because a person checks it. It uses retrieval, generative AI, and human approval. It is fully autonomous.

4. What happens during inference? A trained model uses a new input to make an output. Every training example is labelled. The product name is chosen. The answer is proved correct.

5. Why is a model not the full app? Models cannot use inputs. Apps never use models. Models exist only while they are trained. The app may add rules, search, tools, access control, saved data, safety checks, and people.

6. A spreadsheet copies the latest value into three future months. What do the facts support? Machine learning, because the dates are in the future. Generative AI, because new cells appear. A fixed formula. There is no proof of a trained prediction model. Deep learning, because the file has many sheets.

7. Which system has the clearest signs of an AI agent? A chatbot writes one meeting agenda and stops. A system checks calendars, suggests a time, waits for approval, sends invitations, checks replies, and changes the plan when needed. A system follows one formula. A product has Agent in its name.

8. What does more autonomy mean? The system can do more without a person taking part in every step. The system is more intelligent than every person. Nobody is responsible for its actions. Every action will be correct.

9. `DeepScan Agent` extracts text from forms. No other details are given. What is the best answer? Its name proves deep learning and agent actions. It must use only fixed rules. It cannot be automated. Text extraction is clear. The method and agent claim are not proven.

10. Which is the best simple order for a machine-learning system? Output, product name, input, training, model. Prepare data, train a model, test it, use it on a new input, produce an output. Use the model, collect data, delete the model, produce an output. Write a prompt, get guaranteed truth, approve it automatically.

What to review

Missed 1, 2, or 6: review the system types.
Missed 3 or 5: review the Policy Guide example.
Missed 4 or 10: review training and inference.
Missed 7 or 8: review agents and autonomy.
Missed 9: review TRACE and missing evidence.

Read the feedback before you try again.