Artificial Intelligence

Overview

Rewards

FAQs

Overview

Artificial Intelligence enables machines to learn from data, recognize patterns, make decisions, and solve complex problems with minimal human intervention.
It spans key domains such as machine learning, deep learning, neural networks, natural language processing, reinforcement learning, and ethical AI practices.
The AI Assessment Questionnaire is designed to evaluate your understanding of these core AI concepts and their real-world applications.
It assesses both theoretical foundations and practical awareness across modern AI systems.
By answering these questions, you can benchmark your AI knowledge against industry-relevant standards.

Take this assessment to understand where you stand in Artificial Intelligence and identify areas for further learning.

Read the FAQs tab carefully for Instructions before beginning the assessment.

Rewards

NYXPoints are used to generate the Leaderboard (coming soon). They are awarded for achieving a certain score.

200 nyxpoints for a passing score of 80% or more
300 nyxpoints for a perfect score of 100%
Didn’t pass? You still get 30 nyxpoints for attempting the assesment

FAQs

IMPORTANT instructions for taking the Assessment

The timer starts when you click Start Assessment
DO NOT refresh/reload the page or use the back button to navigate away from the page.
Navigating away from the assessment page DOES NOT stop/pause the timer and the will restart the assessment when you come back to it. The answers are NOT saved.

General

There are NO pre-requisites to take this assessment. Take this assessment even if you are completely new to Linux.
The assessment is completely FREE.
Preferably take it in a closed book mode.
DO NOT copy/paste, share or upload questions elsewhere.

Eligible Rewards

300 NyxCoins*

* NyxCoins vary on score

Questions

Minutes

1 %

Passing Score

Time limit: 0

Assessment Summary

0 of 30 Questions completed

Questions:

Information

You have already completed the assessment before. Hence you can not start it again.

Assessment is loading…

You must sign in or sign up to start the assessment.

You must first complete the following:

Results

Assessment complete. Results are being recorded.

Results

0 of 30 Questions answered correctly

Your time:

Time has elapsed

You have reached 0 of 0 point(s), (0)

Earned Point(s): 0 of 0, (0)
0 Essay(s) Pending (Possible Point(s): 0)

Average score
Your score

Categories

Not categorized 0%

Current
Review
Answered
Correct
Incorrect

Question 1 of 30

1. Question
Which of the following is NOT a limitation of the k-means clustering algorithm?
- Sensitive to initial centroid selection
- Requires predefining the number of clusters (k)
- Works poorly with non-globular clusters
- Guarantees convergence to the global optimum
Correct

Incorrect
Question 2 of 30

2. Question
In reinforcement learning, what does the term “exploration vs. exploitation” refer to?
- Choosing between known rewards and discovering new actions
- Deciding whether to use supervised or unsupervised learning
- Selecting between deep learning and classical ML
- Balancing between training and testing phases
Correct

Incorrect
Question 3 of 30

3. Question
Which activation function is most commonly used in the output layer of a binary classification neural network?
- ReLU
- Sigmoid
- Tanh
- Leaky ReLU
Correct

Incorrect
Question 4 of 30

4. Question
What is the primary purpose of dropout in a neural network?
- To increase training speed
- To reduce overfitting
- To normalize input data
- To optimize learning rate
Correct

Incorrect
Question 5 of 30

5. Question
In Natural Language Processing (NLP), what does BERT stand for?
- Bidirectional Encoder Representations from Transformers
- Binary Encoded Representations for Text
- Basic Embedding of Recurrent Transformers
- Bidirectional Embedding Recognition Technique
Correct

Incorrect
Question 6 of 30

6. Question
Which technique is commonly used to explain individual predictions of complex ML models?
- K-Means
- SHAP
- Dropout
- PCA
Correct

Incorrect
Question 7 of 30

7. Question
What is the main advantage of using a transformer model over RNNs for sequence tasks?
- Lower computational cost
- Better handling of long-range dependencies
- No need for attention mechanisms
- Only works with small datasets
Correct

Incorrect
Question 8 of 30

8. Question
Which of the following is a key challenge in federated learning?
- High computational cost on a single machine
- Privacy preservation across distributed devices
- Requirement for labeled data
- Lack of parallel processing
Correct

Incorrect
Question 9 of 30

9. Question
What does the “No Free Lunch” theorem imply in machine learning?
- No single algorithm works best for all problems
- Supervised learning is always better than unsupervised
- Deep learning outperforms classical ML in every case
- Training models is always computationally expensive
Correct

Incorrect
Question 10 of 30

10. Question
Which optimization algorithm is best suited for noisy or sparse gradients?
- Stochastic Gradient Descent (SGD)
- Adam
- RMSprop
- Momentum
Correct

Incorrect
Question 11 of 30

11. Question
In a GAN (Generative Adversarial Network), what is the role of the discriminator?
- To generate fake data
- To distinguish between real and fake data
- To optimize the generator
- To reduce overfitting
Correct

Incorrect
Question 12 of 30

12. Question
Which of the following is NOT a valid method for handling imbalanced datasets?
- Oversampling the minority class
- Undersampling the majority class
- Using accuracy as the primary metric
- Applying SMOTE (Synthetic Minority Oversampling Technique)
Correct

Incorrect
Question 13 of 30

13. Question
What is the primary purpose of the attention mechanism in neural networks?
- To reduce the number of layers
- To focus on relevant parts of input data
- To speed up backpropagation
- To eliminate vanishing gradients
Correct

Incorrect
Question 14 of 30

14. Question
Which of the following is a key difference between bagging and boosting?
- Bagging reduces variance, while boosting reduces bias
- Bagging uses sequential training, while boosting uses parallel training
- Bagging always outperforms boosting
- Boosting is only used in unsupervised learning
Correct

Incorrect
Question 15 of 30

15. Question
What is the main advantage of using a convolutional neural network (CNN) over a fully connected network for image processing?
- CNNs require fewer parameters due to weight sharing
- CNNs cannot handle large images
- Fully connected networks are always better for images
- CNNs do not use backpropagation
Correct

Incorrect
Question 16 of 30

16. Question
What happens when a machine learning model has high variance?
- It underfits the training data
- It performs well on training but poorly on unseen data
- It has very few parameters
- It cannot learn nonlinear patterns
Correct

Incorrect
Question 17 of 30

17. Question
What is the key idea behind transfer learning?
- Training a model from scratch on a new task
- Using a pre-trained model as a starting point for a related task
- Only using unsupervised learning techniques
- Ignoring previous knowledge when learning new tasks
Correct

Incorrect
Question 18 of 30

18. Question
Which of the following is a disadvantage of using decision trees without pruning?
- They are too interpretable
- They tend to overfit the training data
- They cannot handle categorical features
- They always have low bias
Correct

Incorrect
Question 19 of 30

19. Question
What is data leakage in machine learning?
- When the dataset is too large
- When labels are missing
- When test data information influences model training
- When the learning rate is too high
Correct

Incorrect
Question 20 of 30

20. Question
Which of the following is NOT a type of unsupervised learning?
- K-Means Clustering
- Principal Component Analysis (PCA)
- Autoencoders
- Linear Regression
Correct

Incorrect
Question 21 of 30

21. Question
What is the main challenge in training very deep neural networks?
- Vanishing/exploding gradients
- Too few parameters
- Lack of activation functions
- Overly simple feature extraction
Correct

Incorrect
Question 22 of 30

22. Question
Which of the following is a key feature of a variational autoencoder (VAE)?
- It generates discrete outputs only
- It learns a probabilistic latent space
- It cannot be used for generative tasks
- It does not use backpropagation
Correct

Incorrect
Question 23 of 30

23. Question
What is the primary purpose of the bias term in a neural network?
- To shift the activation function output
- To increase model complexity
- To reduce the number of weights
- To prevent overfitting
Correct

Incorrect
Question 24 of 30

24. Question
Which of the following is a key ethical concern in AI development?
- Model interpretability and fairness
- Using too much training data
- High computational efficiency
- Lack of deep learning models
Correct

Incorrect
Question 25 of 30

25. Question
What is the key advantage of using a recurrent neural network (RNN) over a feedforward network for sequential data?
- RNNs have memory to retain past information
- RNNs cannot handle sequences
- RNNs always train faster
- RNNs do not require backpropagation
Correct

Incorrect
Question 26 of 30

26. Question
Which of the following is NOT a common loss function in deep learning?
- Cross-Entropy Loss
- Mean Squared Error (MSE)
- Hinge Loss
- K-Means Loss
Correct

Incorrect
Question 27 of 30

27. Question
What is the primary purpose of batch normalization in neural networks?
- To reduce internal covariate shift
- To increase training time
- To remove all biases
- To eliminate the need for activation functions
Correct

Incorrect
Question 28 of 30

28. Question
Which of the following is a key difference between supervised and unsupervised learning?
- Supervised learning uses labeled data, while unsupervised does not
- Unsupervised learning always performs better
- Supervised learning cannot be used for classification
- Unsupervised learning requires human feedback
Correct

Incorrect
Question 29 of 30

29. Question
What is the primary goal of the t-SNE algorithm?
- Dimensionality reduction for visualization
- Increasing model complexity
- Optimizing hyperparameters
- Speeding up backpropagation
Correct

Incorrect
Question 30 of 30

30. Question
Which of the following is NOT a valid application of reinforcement learning?
- Game playing (e.g., AlphaGo)
- Robotics control
- Image classification
- Autonomous driving
Correct

Incorrect

Cookie	Duration	Description
cookielawinfo-checkbox-analytics	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Analytics".
cookielawinfo-checkbox-functional	11 months	The cookie is set by GDPR cookie consent to record the user consent for the cookies in the category "Functional".
cookielawinfo-checkbox-necessary	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookies is used to store the user consent for the cookies in the category "Necessary".
cookielawinfo-checkbox-others	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Other.
cookielawinfo-checkbox-performance	11 months	This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Performance".
viewed_cookie_policy	11 months	The cookie is set by the GDPR Cookie Consent plugin and is used to store whether or not user has consented to the use of cookies. It does not store any personal data.