Artificial Intelligence Assignment - Questions and Answers
1(a). Describe Some famous AI systems.
Some famous AI systems include:
- IBM Watson: Known for winning Jeopardy!, used in healthcare.
- AlphaGo: Developed by DeepMind, beat the world champion in Go.
- Siri, Alexa, Google Assistant: Voice-based virtual assistants.
- GPT (like ChatGPT): Large language models for conversation and content generation.
- Tesla Autopilot: AI-based self-driving car system.
1(b). Explain History of AI.
The history of AI started in the 1950s. Alan Turing proposed the Turing Test. In 1956, the
Dartmouth Conference marked the beginning of AI as a field. Early successes included
symbolic reasoning and logic-based systems. The 1980s saw expert systems, and the 2010s
brought deep learning, enabling breakthroughs in vision, speech, and language.
2. Specify a core set of condition - action rules for a VacBot Agent.
- If dirt is present, then suck.
- If bump, then turn right.
- If clean and no bump, move forward.
- If battery low, go to charging station.
3. Express the Wumpus world problem with propositional logic and give the
step-by-step approach using symbolic form.
Let Pij = Pit in cell (i,j), B = Breeze, S = Stench
Rules:
- B11 <-> (P12 ∨ P21)
- S11 <-> (W12 ∨ W21)
Step-by-step:
1. Observe percepts.
2. Translate to logic.
3. Use inference rules to deduce safe/unsafe cells.
4. Compare all search techniques based on Algo, space, time, complete,
optimal.
Breadth First Search: Complete, Optimal (if cost=1), Space: O(b^d), Time: O(b^d)
Depth First Search: Not complete, Not optimal, Space: O(bd), Time: O(b^d)
A*: Complete and optimal (if heuristic is admissible), Space and Time: O(b^d)
Greedy: Not optimal, not complete, Space and Time: O(b^d)
5(a). I am either fat or thin. I’m certainly not thin.
Premises: F ∨ T, ¬T ⇒ Conclusion: F (Modus Tollens and Disjunctive Syllogism)
5(b). If I run I got out of breath. I’m not out of breath.
Premises: R → B, ¬B ⇒ Conclusion: ¬R (Modus Tollens)
5(c). If the butler did it, then his hands are dirty. The butler’s hands are dirty.
Premises: B → D, D ⇒ No conclusion via Modus Ponens; fallacy of affirming the consequent.
5(d). Blue skies make me happy and gray skies make me sad. The sky is either
blue or gray.
Conclusion depends on the current state; multiple possible outcomes. Need more data.
6(a). If my program runs, then I am happy. If I am happy, the sun shines. It’s
11.00 p.m. and very dark.
P → H, H → S, ¬S ⇒ ¬H ⇒ ¬P (Modus Tollens chain)
6(b). All trigonometric functions are periodic functions and all periodic
functions are continuous functions.
T(x) → P(x), P(x) → C(x) ⇒ T(x) → C(x) (Hypothetical Syllogism)
7. Determine whether the following argument is valid or not.
Valid. The chain of logic with disjunction and conditional reasoning supports the conclusion.
8. What are the different chaining rules in logic in AI?
Forward chaining: From facts to conclusion.
Backward chaining: From goal to facts.
Mixed chaining: Combination of both.
9. What is the objective of expert systems? Write down 3 popular expert
systems and explain their design principle.
Objective: Mimic expert-level decision making.
Examples:
1. MYCIN - Medical diagnosis using rule-based system.
2. DENDRAL - Chemical analysis using rule inference.
3. XCON - Configuring computer systems.
Design: Knowledge base, inference engine, user interface.
10. Discuss Bayesian Network with a suitable example.
A Bayesian Network is a probabilistic graphical model representing variables and their
dependencies. Example: Alarm system.
Variables: Burglary (B), Earthquake (E), Alarm (A), JohnCalls (J), MaryCalls (M)
Dependencies: P(A|B,E), P(J|A), P(M|A), etc.
11. We have a database describing 100 examples of printer failures... Show
your work.
Total = 100
Hardware failures = 75 (15 had Windows) ⇒ 15/75 = 0.2
Driver failures = 25 (15 had Windows) ⇒ 15/25 = 0.6
Overall Windows = (15+15)/100 = 0.3 or 30%
Conclusion: 30% of failures involved Windows OS.
12. Probability you have disease given positive test.
P(Disease) = 0.0001
P(¬Disease) = 0.9999
P(Pos | Disease) = 0.99
P(Pos | ¬Disease) = 0.01
Using Bayes' Theorem:
P(D | Pos) = [0.99 * 0.0001] / [0.99*0.0001 + 0.01*0.9999] ≈ 0.0098 ⇒ ~0.98%
13(a). Coin toss - probability fake coin after head.
P(Fake | Head) = [P(Head | Fake) * P(Fake)] / Total
= (1 * 1/11) / (1/11 + 0.5*10/11) = (1/11) / (1/11 + 5/11) = 1/6 ≈ 0.167
13(b). After repeated heads, probability fake?
Let n heads observed. P(Fake | n heads) = (1 * 1/11) / [1/11 + (0.5^n * 10/11)] ⇒ increases
with n.
E.g., for n = 3, P = 1 / (1 + 10/8) ≈ 0.444
14. Steel plant production defect - what’s the probability of source boiler?
Let A = product from B1 (30%, 2%), B2 (30%, 3%), B3 (40%, 4%)
P(Def) = 0.3*0.02 + 0.3*0.03 + 0.4*0.04 = 0.031
P(B1|D) = 0.3*0.02/0.031 ≈ 0.1935
P(B2|D) = 0.3*0.03/0.031 ≈ 0.2903
P(B3|D) = 0.4*0.04/0.031 ≈ 0.5161
15. Biased coin – probability first head on even toss.
Let p = probability of head.
Sample space = sequences like T, TH, TTH...
P(Even) = (1-p)*p + (1-p)^3*p + (1-p)^5*p + ... = p*(1-p)*[1 + (1-p)^2 + (1-p)^4 + ...] =
(p*(1-p)) / (1 - (1-p)^2)
= p*(1-p) / (2p - p^2) = (1 - p) / (2 - p)
