- LPU MCA Semester 1
- CAB107
- Unit 1
CAB107 · Unit 1: Foundations of AI and data science
Evolution and industry applications of AI
10 practice questions · in the MTE and ETE syllabus
Key points
- Turing test: a judge cannot tell machine from human in conversation
- AI winters were funding collapses after unmet promises
- Deep learning boom came from big data and GPUs
- Deep Blue (chess, 1997) differs from AlphaGo (Go, 2016)
AI has developed in waves: periods of optimism followed by AI winters, when funding and interest fell because results did not match promises.
| Era | Milestone |
|---|---|
| 1950 | Turing test proposed: can a machine converse so that a human judge cannot tell it from a human |
| 1956 | Dartmouth workshop names the field |
| 1960s | ELIZA chatbot |
| 1980s | Expert systems boom, then costly to maintain and a second winter |
| 1997 | Deep Blue beats chess champion Garry Kasparov |
| 2010s | Deep learning boom from large datasets plus GPU computing |
| 2016 | AlphaGo beats Go champion Lee Sedol |
Expert systems became costly because every rule was written and updated by hand by knowledge engineers, and the rule bases grew brittle as they scaled. The deep learning boom was driven by big data, GPUs and better algorithms.
Typical industry applications include recommendation systems in retail, fraud detection in banking, medical image analysis in healthcare, speech recognition and machine translation in language technology, and perception in autonomous driving. Real solutions often combine capabilities, for example computer vision to read scanned forms and language processing to interpret the text.
Exam trap: match milestones to the right system. Deep Blue played chess in 1997; AlphaGo played Go in 2016.
Try two questions
What is an AI winter?
- A period when funding and interest in AI fell because the field failed to meet its promises
- A season in which data centres lower their cooling costs by making use of the cold winter weather outside
- A method for removing old and unused records from a training set to save storage space
- The first stage of training a neural network, in which weights start at random values
Show the answer
A. Over-promising and limits of early hardware and methods led to reduced funding in the 1970s and late 1980s.
Why not B: It is not a cooling technique; the term is about funding and enthusiasm.
Why not C: AI winter is not a data-cleaning procedure.
Why not D: It is a historical phase of the field, not a training stage.
A claims-processing firm wants to replace manual reading of scanned forms. Which AI capability combination is required?
- Reinforcement learning for playing board games and a recommender system for suggesting ads
- Computer vision for reading the scans and language processing for interpreting the text
- Speech synthesis for producing spoken replies and robotics for moving boxes in a warehouse
- Time-series forecasting of weather and clustering of galaxies by their brightness
Show the answer
B. Scanned forms need OCR style vision first and then text understanding.
Why not A: Neither game playing nor ad recommendation reads forms.
Why not C: Neither speaking nor moving objects reads documents.
Why not D: Neither weather forecasting nor galaxy clustering reads scanned forms.