- LPU MCA Semester 1
- CAB107
- Unit 1
CAB107 · Unit 1: Foundations of AI and data science
AI, machine learning, deep learning and data science: definitions and differences
10 practice questions · in the MTE and ETE syllabus
Key points
- AI contains ML, and ML contains deep learning
- ML infers rules from data instead of being given them
- Data science covers the whole path from data to insight
- Rule-based AI is not machine learning
Artificial intelligence (AI) is the broad field of building systems that perform tasks normally needing human intelligence. Machine learning (ML) is a subset of AI in which a system learns patterns from example data instead of following hand-written rules. Deep learning (DL) is a subset of ML that uses neural networks with many layers. Data science is the wider practice of extracting insight from data, covering collection, cleaning, analysis, modelling and communication; ML is one tool inside it.
| Term | Scope | Example |
|---|---|---|
| AI | Widest | A rule-based chess program |
| ML | Inside AI | Spam filter trained on labelled mail |
| DL | Inside ML | CNN recognising objects in photos |
| Data science | Overlaps all | Sales analysis ending in a dashboard |
In traditional programming the programmer supplies rules plus data and gets answers. In ML the programmer supplies data plus answers and the algorithm infers the rules. Deep learning usually needs more data and computing power, typically GPUs, but it learns its own features from raw input such as pixels.
Narrow AI performs one task well; artificial general intelligence (AGI) with human-level breadth does not yet exist.
Exam trap: AI, ML and DL are nested, not synonyms. A fixed list of if-else rules can be AI but is not ML, because nothing is learned from data.
Try two questions
Which statement best separates data science from machine learning?
- Data science is a subfield of deep learning that deals only with image and video data
- Machine learning covers data collection, cleaning and reporting, and data science covers only the step of training models
- The two terms mean exactly the same thing and can be swapped freely in every situation
- Data science covers the whole path from collecting data to communicating insight, and ML is one modelling tool inside it
Show the answer
D. Data science is an end-to-end discipline using statistics, programming and domain knowledge; ML is one technique it may use.
Why not A: Data science is far broader than deep learning and works on many data types.
Why not B: This reverses the scope; model training is the ML part of the wider data science job.
Why not C: They overlap but are not identical; a dashboard analysis can be data science without any ML.
A bank has 5 million past transactions each labelled fraud or genuine. Which approach fits the goal of flagging new fraud?
- Train a machine learning classifier on the labelled transactions
- Sort the transactions by amount and flag the top row
- Compute the average transaction amount per month in a pivot table
- Write one fixed rule that never changes and never uses the labels
Show the answer
A. Labelled history is exactly the input a supervised classifier needs.
Why not B: Sorting does not learn any pattern separating fraud from genuine cases.
Why not C: A monthly average is a summary and cannot classify a new transaction.
Why not D: A frozen rule ignores the labelled history that supervised learning is meant to exploit.