(2nd Sem) AIML Unit 2: Questions & Answers
Unit – II: Intelligent Agents -> Generated and Prepared By Thiruselvan (ThiruXD)
A. Multiple Choice Questions (with Answers)
-
An intelligent agent perceives its environment through:
a) Actuators b) Sensors c) Goals d) Rules
Answer: b) Sensors
-
The mathematical representation of an agent function is:
a) f: A → P* b) f: P* → A c) f: E → A d) f: S → G
Answer: b) f: P → A*
-
Which characteristic means the agent operates without direct human intervention?
a) Reactivity b) Autonomy c) Social Ability d) Rationality
Answer: b) Autonomy
-
Simple reflex agents act only on:
a) Past history b) Current percept c) Future goals d) Utility function
Answer: b) Current percept
-
Which type of agent maintains an internal model of the world?
a) Simple reflex b) Model-based c) Goal-based d) Utility-based
Answer: b) Model-based
-
Goal-based agents select actions by considering:
a) Only current percept b) Future states relative to goals c) Only utility d) Random choice
Answer: b) Future states relative to goals
-
Utility-based agents use a utility function to:
a) Store history b) Quantify how desirable an outcome is c) Follow fixed rules d) Map rooms
Answer: b) Quantify how desirable an outcome is
-
PEAS stands for:
a) Performance, Environment, Actuators, Sensors
b) Perception, Evaluation, Action, State
c) Planning, Environment, Agents, Sensors
d) Performance, Evaluation, Actuators, State
Answer: a) Performance, Environment, Actuators, Sensors
-
Chess is an example of which type of environment?
a) Partially observable & Stochastic b) Fully observable & Deterministic
c) Dynamic & Continuous d) Partially observable & Dynamic
Answer: b) Fully observable & Deterministic
-
A self-driving car operates in a:
a) Fully observable, Static, Discrete environment
b) Partially observable, Dynamic, Continuous environment
c) Fully observable, Deterministic, Static environment
d) Discrete, Static, Deterministic environment
Answer: b) Partially observable, Dynamic, Continuous environment
-
A rational agent always tries to:
a) Follow fixed rules b) Maximize its performance measure
c) Minimize sensors d) Ignore uncertainty
Answer: b) Maximize its performance measure
-
Which agent type is best for making trade-offs between multiple objectives?
a) Simple reflex b) Model-based c) Goal-based d) Utility-based
Answer: d) Utility-based
-
Roomba is an example of:
a) Simple reflex agent only b) Model-based agent
c) Pure software agent d) Utility-based financial agent
Answer: b) Model-based agent
-
In PEAS, “Actuators” refer to:
a) Devices that sense the environment b) Mechanisms that perform actions
c) Criteria for success d) The world of the agent
Answer: b) Mechanisms that perform actions
-
An environment where the outcome of an action is always the same is called:
a) Stochastic b) Deterministic c) Dynamic d) Continuous
Answer: b) Deterministic
-
Which environment changes while the agent is deciding?
a) Static b) Dynamic c) Discrete d) Fully observable
Answer: b) Dynamic
-
Performance measure is:
a) Internal to the agent b) Set by the designer and external to the agent
c) Always the number of actions d) Same for all agents
Answer: b) Set by the designer and external to the agent
-
Chatbots and recommendation systems are examples of:
a) Autonomous robots b) Software agents c) Simple reflex only d) Physical actuators
Answer: b) Software agents
-
Which agent has no memory and no learning capability?
a) Model-based b) Goal-based c) Simple reflex d) Utility-based
Answer: c) Simple reflex
-
“Human-in-the-loop” is most related to which characteristic?
a) Autonomy b) Rationality c) Social Ability d) Proactiveness
Answer: a) Autonomy (agents still often need human oversight in critical systems)
B. Theory Questions with Answers
1. Define an Intelligent Agent. Explain its four key components.
Answer:
An intelligent agent is an autonomous entity that perceives its environment through sensors, processes information to reason, and takes action via actuators to achieve specific goals. It operates continuously and can improve with experience.
Four key components:
- Perception – Uses sensors to observe the environment.
- Reasoning – Processes the perceived information using algorithms, logic or learning.
- Decision-Making – Chooses actions based on goals or objectives.
- Action – Executes actions through actuators to change the environment.
2. What is the Agent Function? Write its mathematical form.
Answer:
An agent is a system that maps percept sequences (history of everything perceived) to actions.
Mathematically:
f : P\ → A*
where P* = sequence of percepts, A = action taken.
3. List and explain the five main characteristics of Intelligent Agents.
Answer:
- Autonomy – Operates without direct human intervention; controls its own behaviour.
- Reactivity – Perceives the environment and responds timely to changes.
- Proactiveness (Goal-orientation) – Takes initiative to achieve goals, not just reacts.
- Social Ability – Can interact with other agents or humans using communication.
- Rationality – Acts to maximize its performance measure given the percepts and knowledge.
4. Explain Simple Reflex Agents with characteristics and example.
Answer:
Simple reflex agents select actions based only on the current percept using condition-action rules. They have no memory of past percepts.
Characteristics:
- Purely reactive
- Fast response
- No learning
- Suitable only for fully observable and stable environments
Example: Traffic light system that changes based on fixed timing rules.
5. Differentiate between Model-based and Goal-based agents.
Answer:
| Feature | Model-based Agent | Goal-based Agent |
|---|---|---|
| Main feature | Maintains internal state/model of world | Has explicit goals and plans actions |
| Decision basis | Current percept + internal model | Future states that lead to goal |
| Capability | Handles partial observability | Can plan sequences of actions |
| Flexibility | Updates model when environment changes | Can re-plan if conditions change |
| Example | Roomba (maps room & tracks cleaned areas) | Logistics routing agent |
6. What are Utility-based Agents? Why are they more advanced?
Answer:
Utility-based agents extend goal-based agents by using a utility function that measures how desirable a particular state or outcome is.
They can make trade-offs between multiple conflicting goals (cost, time, risk, quality, etc.) and choose the action that gives the highest expected utility. This makes them suitable for uncertain and complex environments where simply reaching a goal is not enough — the quality of the goal achievement also matters.
Example: Financial portfolio management agent that balances risk and return.
7. Explain the PEAS framework with an example.
Answer:
PEAS is a structured way to describe any AI agent:
- P – Performance Measure: Criteria for success (e.g., cleanliness, safety, accuracy)
- E – Environment: The world in which the agent operates
- A – Actuators: Mechanisms to perform actions
- S – Sensors: Devices to perceive the environment
Example – Roomba:
- Performance: Area cleaned, dirt collected, battery life, no collisions
- Environment: Rooms, furniture, floor, obstacles
- Actuators: Wheels, vacuum motor, brushes
- Sensors: Bumper sensors, cliff sensors, dirt sensors, camera/mapping sensors
8. Compare Fully Observable vs Partially Observable environments.
Answer:
| Aspect | Fully Observable | Partially Observable |
|---|---|---|
| Information | Agent gets complete state | Agent gets incomplete/limited information |
| Hidden states | None | Some states are hidden |
| Decision making | Simpler | Needs memory or estimation |
| Example | Chess | Self-driving car |
9. Explain Deterministic vs Stochastic and Static vs Dynamic environments.
Answer:
Deterministic vs Stochastic:
- Deterministic → Same action always produces same result (e.g., puzzle).
- Stochastic → Outcome has randomness/uncertainty (e.g., weather, traffic).
Static vs Dynamic:
- Static → Environment does not change while agent is deciding (e.g., crossword).
- Dynamic → Environment keeps changing (e.g., driving a car).
10. What is Rationality? How is it related to Performance Measure?
Answer:
A rational agent is one that selects actions expected to maximize its performance measure, based on the percept sequence and its built-in knowledge.
Performance measure is defined by the designer and is external to the agent. Rationality means the agent always chooses the action that gives the best expected score according to that measure (considering uncertainty if present).
C. Analytical / Application Questions with Answers
1. Classify the following agents and justify:
- Traffic light controller
- Roomba
- Self-driving car
- Netflix recommendation system
Answer:
- Traffic light → Simple Reflex (fixed condition-action rules, no memory).
- Roomba → Model-based (maintains map of room and cleaned areas).
- Self-driving car → Utility-based / Goal-based (has goals of safety + destination, makes complex trade-offs).
- Netflix → Utility-based software agent (maximizes user satisfaction/utility of recommendations).
2. Give PEAS description for a Self-driving Car.
Answer:
- Performance Measure: Safety, reaching destination, passenger comfort, fuel/energy efficiency, traffic rule compliance.
- Environment: Roads, traffic, pedestrians, weather, traffic signals.
- Actuators: Steering, accelerator, brakes, indicators, horn.
- Sensors: Cameras, LiDAR, Radar, GPS, ultrasonic sensors, speed sensors.
3. Why can’t a Simple Reflex agent work well for a self-driving car? Which type is needed and why?
Answer:
Simple reflex agents have no memory and act only on current percept. A self-driving car needs history (previous positions, speed, intentions of other vehicles) and must handle partial observability.
It needs at least a Model-based agent (to maintain internal state) and preferably Goal-based or Utility-based agents (to plan routes and make safety vs speed trade-offs).
4. A vacuum agent has two locations (A and B). Dirt may appear randomly. Which environment properties does it have? Which agent type is suitable?
Answer:
- Partially observable (cannot see both rooms at once unless it has memory)
- Stochastic (dirt appears randomly)
- Dynamic (dirt can appear while cleaning)
- Discrete
Suitable agent: Model-based reflex agent (maintains which rooms are clean) or Goal-based agent.
5. Differentiate Goal-based and Utility-based agents with a real-life example.
Answer:
A Goal-based delivery robot only cares about reaching the destination.
A Utility-based delivery robot also considers fuel cost, time, traffic risk, and package safety, and chooses the route that maximizes overall utility (best compromise).
Utility-based agents are more powerful because they can handle multiple conflicting objectives.
6. Explain how rationality is evaluated in an intelligent agent.
Answer:
Rationality is evaluated using the performance measure defined by the designer.
The agent is rational if it chooses actions that maximize the expected performance measure, given:
- What it has perceived so far
- Its knowledge of the environment
- The available actions
Evaluation can be quantitative (accuracy, time, cost), qualitative (user satisfaction), or comparative (vs other agents).