🧠 Arti cial Intelligence Notes
✅ Types of AI
1. Narrow AI (Weak AI)
◦ Performs a speci c task (e.g., Siri, Google Translate).
2. General AI (Strong AI)
◦ Mimics human intelligence across any task (not achieved yet).
3. Super AI
◦ Hypothetical AI that surpasses human intelligence.
✅ AI Domains
1. Natural Language Processing (NLP)
◦ Enables machines to understand human language.
◦ Example: Chatbots, translators.
2. Computer Vision
◦ Enables machines to interpret visual data.
◦ Example: Facial recognition, object detection.
3. Expert Systems
◦ Uses knowledge base and rules to solve problems.
◦ Example: Medical diagnosis systems.
4. Robotics
◦ Combines AI with mechanical automation.
◦ Example: Industrial robots.
5. Machine Learning
◦ Systems learn patterns from data.
◦ Example: Email spam ltering.
✅ AI Project Cycle
fi
fi
fi
1. Problem Scoping
◦ De ne the problem clearly.
◦ Example: Predict student dropout.
2. Data Acquisition
◦ Collect relevant data.
◦ Example: Attendance, marks, behavior.
3. Data Exploration
◦ Analyze the data using charts/statistics.
4. Modelling
◦ Use algorithms to train AI.
◦ Example: Decision Trees, SVMs.
5. Evaluation
◦ Test model accuracy and improve.
✅ 4Ws Canvas
Elemen
Description Example
t
What What is the problem? Students dropping out
Affects future education
Why Why is the problem important?
outcomes
Who Who are the stakeholders? Students, teachers, parents
Where will the solution be
Where Schools, colleges
used?
✅ Ethics of AI
1. Fairness – No bias against any group.
2. Transparency – Model decisions should be understandable.
3. Accountability – People must be responsible for AI outcomes.
4. Privacy – Respect data ownership and user privacy.
5. Security – AI should not be easily hacked or manipulated.
fi
✅ System Map
Root Node: Main goal or issue.
• Example: Improve public safety.
Branches/Sub-nodes:
• CCTV AI monitoring
• AI emergency response
• Traf c violation detection
✅ Rule-Based vs Learning-Based AI
Feature Rule-Based AI Learning-Based AI
Uses “if-then”
Logic Learns patterns from data
rules
Machine Learning
Example Expert Systems
models
Adaptabilit
Rigid Can adapt
y
Data Need Low High
✅ Difference Between Ethics and Morality
Ethics Morality
Social/system-based Personal belief-based
Can differ by
More universal
context
Example: AI Example: Telling the
fairness truth
✅ Data
• De nition: Raw facts and gures (text, numbers, images, etc.)
• Types:
◦ Structured (tables)
◦ Unstructured (images, audio)
fi
fi
fi
• Use in AI: Training, testing, evaluating models
✅ Statistical Data Concepts
• Mean, Median, Mode – Central tendency
• Variance/Standard Deviation – Spread of data
• Correlation – Relationship between variables
✅ Computer Vision
• De nition: Allows machines to interpret images or videos.
• Applications:
◦ Face detection
◦ OCR (Optical Character Recognition)
◦ Self-driving cars
✅ Natural Language Processing (NLP)
• De nition: Enables machines to process and understand human language.
• Tasks:
◦ Sentiment Analysis
◦ Translation
◦ Chatbots
◦ Speech recognition
📝 50 MCQs – Practice Questions
1. What type of AI is Siri?
A) General AI B) Super AI C) Narrow AI D) None
✅ C) Narrow AI
2. Which AI domain focuses on interpreting images?
A) NLP B) Computer Vision C) Robotics D) Planning
✅ B) Computer Vision
fi
fi
3. What is the rst stage in the AI project cycle?
A) Modelling B) Evaluation C) Problem Scoping D) Data Collection
✅ C) Problem Scoping
4. Which of these is a rule-based system?
A) Deep Learning B) Decision Tree C) Expert System D) CNN
✅ C) Expert System
5. NLP stands for:
A) Neural Learning Process
B) Natural Language Processing
C) New Logic Programming
D) None of the above
✅ B) Natural Language Processing
6. Morality is more:
A) Law-based B) Society-based C) Personal-based D) Political
✅ C) Personal-based
7. Which AI type is purely hypothetical?
A) Weak AI B) Strong AI C) Super AI D) Domain AI
✅ C) Super AI
8. What does the “Why” in 4Ws canvas represent?
A) Stakeholder
B) Importance of problem
C) Data source
D) Tools used
✅ B) Importance of problem
9. Computer vision is commonly used in:
A) Translation B) Speech Recognition C) Face Recognition D) Grammar Check
✅ C) Face Recognition
10. Data analysis is part of which AI project stage?
A) Modelling B) Evaluation C) Exploration D) Deployment
✅ C) Exploration
11. Which of these is NOT an AI domain?
A) Robotics
B) Computer Vision
C) Thermodynamics
D) NLP
✅ C) Thermodynamics
12. In ethics, which principle ensures AI treats all users fairly?
A) Accuracy
B) Fairness
C) Privacy
fi
D) Security
✅ B) Fairness
13. Which term refers to machines learning from data without explicit programming?
A) Rule-based AI
B) Super AI
C) Machine Learning
D) Vision AI
✅ C) Machine Learning
14. Data collected from sensors is usually:
A) Structured
B) Tabular
C) Unstructured
D) Fictional
✅ C) Unstructured
15. A face recognition system is part of:
A) NLP
B) Computer Vision
C) Speech Processing
D) Rule-based AI
✅ B) Computer Vision
16. "If-then" logic is used in:
A) Neural Networks
B) Rule-Based Systems
C) Deep Learning
D) NLP
✅ B) Rule-Based Systems
17. Which AI type is designed for general problem solving like humans?
A) Narrow AI
B) Strong AI
C) Weak AI
D) Reactive AI
✅ B) Strong AI
18. A chatbot is an example of:
A) NLP
B) Robotics
C) Statistics
D) Planning
✅ A) NLP
19. Which of the following is a part of data exploration?
A) Building Models
B) Plotting Graphs
C) Installing Software
D) Final Evaluation
✅ B) Plotting Graphs
20. AI can be biased if:
A) Data is cleaned
B) Data is balanced
C) Data has prejudice
D) Data is large
✅ C) Data has prejudice
21. Rule-based systems depend on:
A) Human-created logic
B) Neural networks
C) Data lakes
D) Sensors
✅ A) Human-created logic
22. Root node in system mapping indicates:
A) Cause
B) Sub-solution
C) Main goal
D) Result
✅ C) Main goal
23. Ethics are based on:
A) Individual beliefs
B) Personal choice
C) Social norms and systems
D) Religion only
✅ C) Social norms and systems
24. NLP does NOT include:
A) Speech Recognition
B) Text Generation
C) Image Captioning
D) Sentiment Analysis
✅ C) Image Captioning
25. Mean, median, and mode are types of:
A) AI models
B) Statistics
C) Algorithms
D) Programming
✅ B) Statistics
26. Which of the following is used in supervised learning?
A) Labeled Data
B) Unlabeled Data
C) No Data
D) Pretrained Data
✅ A) Labeled Data
27. AI is different from traditional programming because it:
A) Is always accurate
B) Learns from data
C) Needs no data
D) Doesn’t use logic
✅ B) Learns from data
28. An AI that beats a human at chess is an example of:
A) Weak AI
B) Strong AI
C) Super AI
D) General AI
✅ A) Weak AI
29. Learning-based AI is:
A) Static
B) Dynamic
C) Prede ned
D) Manual
✅ B) Dynamic
30. Computer vision allows machines to:
A) Hear
B) Speak
C) Read text only
D) Understand images and videos
✅ D) Understand images and videos
31. Which stage of the AI cycle involves checking model accuracy?
A) Data Collection
B) Exploration
C) Evaluation
D) Scoping
✅ C) Evaluation
32. Which of these is a machine learning algorithm?
A) Python
B) Excel
C) Decision Tree
D) Google Sheets
✅ C) Decision Tree
33. If a model performs well only on training data, it's called:
A) Under tting
B) Normal
C) Over tting
fi
fi
fi
D) Generalized
✅ C) Over tting
34. Stakeholders in AI projects include:
A) Only developers
B) Only end users
C) Anyone affected by the solution
D) Only managers
✅ C) Anyone affected by the solution
35. Privacy in AI means:
A) Always showing data
B) Hiding algorithms
C) Protecting personal data
D) Using less data
✅ C) Protecting personal data
36. A machine understanding spoken commands is doing:
A) Image processing
B) Text mining
C) Speech recognition
D) Classi cation
✅ C) Speech recognition
37. Unstructured data includes:
A) Tables
B) Excel les
C) Audio clips
D) SQL records
✅ C) Audio clips
38. Which approach adapts based on new data?
A) Rule-Based
B) Static Models
C) Learning-Based
D) Manual Programming
✅ C) Learning-Based
39. Data cleaning is important because:
A) It reduces data
B) It improves model performance
C) It deletes information
D) It stops data collection
✅ B) It improves model performance
40. Computer vision is mainly used in:
A) Grammar checkers
B) Object detection
C) Music creation
fi
fi
fi
D) Autocomplete
✅ B) Object detection
41. In the 4Ws canvas, “Where” refers to:
A) Data Source
B) Target location of solution
C) Tool used
D) Time spent
✅ B) Target location of solution
42. Which system mimics human decision-making through rules?
A) Neural Net
B) Deep Learning
C) Expert System
D) CNN
✅ C) Expert System
43. AI in healthcare helps in:
A) Fixing machines
B) Diagnosing diseases
C) Operating banks
D) Writing poems
✅ B) Diagnosing diseases
44. Statistical term for data spread:
A) Mean
B) Variance
C) Median
D) Mode
✅ B) Variance
45. A self-driving car uses which AI domains?
A) NLP only
B) Robotics & CV
C) Rule-Based AI only
D) None
✅ B) Robotics & CV
46. A machine translating French to English uses:
A) Speech-to-text
B) NLP
C) Computer Vision
D) Expert Systems
✅ B) NLP
47. A highly intelligent, conscious AI is:
A) Narrow AI
B) Super AI
C) Rule-Based AI
D) Reactive Machine
✅ B) Super AI
48. Which is more exible?
A) Rule-Based AI
B) Hardcoded Programs
C) Learning-Based AI
D) Static Systems
✅ C) Learning-Based AI
49. AI project cycle ends with:
A) Data collection
B) Evaluation
C) Modelling
D) Exploration
✅ B) Evaluation
50. AI fairness ensures:
A) Everyone gets equal outcomes
B) No group is unfairly treated
C) Data is reduced
D) Systems crash less
✅ B) No group is unfairly treated
fl
