Basic Artificial Intelligence Questions

1. What is Artificial Intelligence (AI)?

Making computers smart so they can think and learn like humans. Examples: Recognizing faces, driving cars, or playing chess.

2. What are the different types of AI?

  • Weak AI (Narrow AI): Good at one specific job (e.g., Siri, Alexa, playing Chess).
  • Strong AI (General AI): Smart like a human (can handle any task). Does not exist yet.
  • Super AI: Smarter than humans (Sci-fi movie stuff).

3. What is the difference between AI, Machine Learning (ML), and Deep Learning (DL)? Think of Russian dolls:

  • AI: The big outer shell (The concept of smart machines).
  • ML (Machine Learning): Inside AI. Machines learning from data instead of strict rules.
  • DL (Deep Learning): Inside ML. Mimics the human brain to learn complex things.

4. What is the Turing Test?

A test to see if a computer can fool a human into thinking it’s also a human during a conversation.

5. What are Intelligent Agents?

A system that uses sensors (eyes) to see the world and actuators (hands) to act on it to achieve a goal. Like a robot vacuum cleaner.

Intermediate Questions

6. Explain the difference between Supervised and Unsupervised Learning.

  • Supervised Learning: Learning with a Teacher. The data has answers (labels). Example: Teaching a child “This is an Apple”.
  • Unsupervised Learning: Learning without a Teacher. The data has no answers. The machine finds patterns on its own. Example: Grouping similar fruits together without knowing their names.

7. What is Overfitting and Underfitting?

  • Overfitting: Memorizing the study material but failing the exam because you can’t solve new questions. (Model is too complex).
  • Underfitting: Not studying enough. You fail because you didn’t learn the basics. (Model is too simple).

8. What is a Neural Network?

A computer system designed to work like a human brain. It uses layers of “neurons” to process information and find patterns.

9. What is Natural Language Processing (NLP)?

Teaching computers to understand and speak human language. Examples: Google Translate, Siri, Chatbots.

10. What is the confusion matrix?

A scorecard for your AI model. It shows where the model guessed right and where it got confused (made mistakes).

11. What is Gradient Descent?

An algorithm to minimize errors. Think of it like walking down a mountain blindfolded to find the lowest point (least error) by taking small steps downwards.

Advanced Questions

12. What is Backpropagation?

The way Neural Networks learn. It looks at the mistake (error) at the end and goes backward through the network to tweak the settings so it doesn’t make the same mistake again.

13. What is the difference between Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN)?

  • CNN (Convolutional): Good for Images (has “eyes”). Used for face recognition.
  • RNN (Recurrent): Good for Sequence/Text (has “memory”). Used for translation or predicting the next word.

14. What is the Vanishing Gradient Problem?

When a network is too deep, the early layers stop learning because the “feedback signal” gets too weak as it travels backward. Like a game of “Chinese Whispers” where the message is lost at the start.

15. What are Generative Adversarial Networks (GANs)?

Two AIs fighting each other in a game:

  • The Faker (Generator): Tries to create fake images.
  • The Cop (Discriminator): Tries to catch the fakes.
  • They both get better over time, resulting in realistic fake images.

16. What is Reinforcement Learning?

Training by Trial and Error. Think of training a dog: give a treat (reward) for good behavior and a scolding (penalty) for bad behavior.

17. What are Transformers in AI?

A modern AI model that pays “Attention” to the important words in a sentence, no matter where they are. It powers technologies like ChatGPT and Google Translate.

18. What is Transfer Learning?

Using knowledge from one task to help with another. Like if you know how to ride a bike, it’s easier to learn how to ride a motorcycle. You don’t start from zero.

19. What is Explainable AI (XAI)?

AI that can explain why it made a decision. Instead of a “Black Box” (magic), it tells you “I rejected the loan because income was low.”

20. What is the difference between Epoch, Batch, and Iteration?

  • Epoch: Reading the whole textbook once.
  • Batch: Reading just one chapter at a time.
  • Iteration: How many chapters you need to read to finish the book.

21. What are Hyperparameters?

The settings knobs you tune before training the model. Like setting the temperature and time on an oven before baking a cake.

22. Explain A Search Algorithm.*

A smart way to find the shortest path on a map (like Google Maps). It guesses which way is promising to reach the destination faster.

23. What is Tokenization in NLP?

Chopping up a sentence into pieces (words or letters) so the computer can digest it. E.g., “I love AI” -> [“I”, “love”, “AI”].

24. What is Regularization?

A technique to prevent Overfitting. It penalizes the model for being too complex, forcing it to keep things simple and general.

25. What is a Decision Tree?

A flowchart-like structure where you answer “Yes” or “No” questions to reach a decision. Like playing a game of “20 Questions”.

26. What is Random Forest?

A collection of many Decision Trees. It’s like asking a crowd of people for their opinion instead of just one person (Wisdom of Crowds) to get a better answer.

27. What is Dimensionality Reduction?

Simplifying data by removing unnecessary details. Like summarizing a long book into a few pages without losing the main story.

28. What is Cross-Validation?

A way to test if your model is reliable. You split your data into different parts and test the model on each part to ensure it works well everywhere, not just on one specific set.

29. What is Computer Vision?

Teaching computers to see and understand images or videos. Used in self-driving cars to spot pedestrians or in phones for face unlock.

30. What is Bias in AI?

When an AI makes unfair decisions because it was trained on bad or incomplete data. Example: A hiring AI that favors one gender because it was mostly trained on resumes from that gender.

31. What is Data Augmentation?

Creating more training data from what you already have. For images, this means flipping, rotating, or zooming in so the AI sees more variations and learns better.

32. What is Big Data?

Data that is so huge and fast that traditional tools can’t handle it. AI needs Big Data to learn effectively.

33. What is Python’s role in AI?

Python is the most popular language for AI because it is easy to read (like English) and has powerful “toolkits” (libraries like TensorFlow and PyTorch) that do the heavy math for you.

34. What is a Perceptron?

The simplest form of a Neural Network. Think of it as a single neuron that takes inputs, weighs them, and decides “Yes” or “No”.

35. What is an Activation Function?

The “Switch” in a neuron that decides if it should fire or not based on the input. Examples: Sigmoid (smooth curve), ReLU (stops negatives). Without it, the network can’t learn complex patterns.

36. What is Dropout?

A way to prevent Overfitting. We randomly “turn off” some neurons during training so the network doesn’t rely too much on any single neuron. It forces the team (neurons) to work better together.

37. What is an Autoencoder?

A Neural Network that learns to compress data into a small code and then recreate it. Like zipping a file and then unzipping it, often used to remove noise from images.

38. What is the difference between Batch and Stochastic Gradient Descent?

  • Batch: Checks all study material before making a correction. Accurate but slow.
  • Stochastic: Checks one item and corrects immediately. Fast but a bit messy/random.

Ethics in AI

39. What is AI Ethics?

Rules and morals to stop AI from hurting people. For example, making sure AI isn’t biased or doesn’t steal private data.

40. What are Deepfakes?

Fake videos or audio made by AI that look 100% real. They are dangerous because they can make people appear to say or do things they never did.

41. Why is Privacy a big issue in AI?

AI needs a lot of data to learn. Sometimes it uses personal information (photos, chats, health info) without permission, which violates privacy.

42. What is the “Black Box” problem?

When an AI is so complex that even its creators don’t understand how it made a specific decision. It’s risky because we can’t trust or fix mistakes we don’t understand.

43. Will AI take away jobs?

It will likely automate boring, repetitive tasks (like data entry), but it will also create new types of jobs. Humans will need to adapt and learn new skills to work alongside AI.