This first set of Artificial Intelligence MCQs builds the foundation: what makes an agent rational, how to specify a task with PEAS, the differences between reflex, model-based, goal-based, utility-based and learning agents, and how to classify environments as observable, deterministic, episodic or dynamic. It also revisits the Turing test and early AI history. The questions suit university semester exams, GATE and UGC NET aspirants, and placement interviews that open with AI fundamentals.
How to practise: read each question and tap the option you think is correct. You will see right away whether you got it, along with a short explanation, and your score is tracked at the bottom of the screen. This set has 18 questions.
Topics covered in this set
- Rational agents and the agent function
- PEAS task specification
- Simple reflex, model-based, goal-based and utility-based agents
- Learning agent components
- Properties of task environments
- Turing test and Chinese Room argument
Artificial Intelligence MCQs with answers and explanations
Q1. In the PEAS framework used to specify a task environment, what does the letter A stand for?
a) Algorithms
b) Actuators
c) Agents
d) Accuracy
Q2. Solving a crossword puzzle alone is best classified as which kind of task environment?
a) Fully observable, deterministic, sequential, static
b) Partially observable, stochastic, episodic, dynamic
c) Fully observable, stochastic, episodic, static
d) Partially observable, deterministic, sequential, semidynamic
Q3. Chess played with a clock is described as which type of environment with respect to time?
a) Static
b) Dynamic
c) Semidynamic
d) Episodic
Q4. A simple reflex agent selects actions only from the current percept. Under which condition is it guaranteed to behave correctly?
a) The environment is stochastic
b) The agent has a utility function
c) The agent stores the full percept history
d) The environment is fully observable
Q5. What extra element does a model-based reflex agent maintain compared with a simple reflex agent?
a) An internal state that tracks aspects of the world it cannot currently observe
b) A search tree of future action sequences
c) A numeric utility for every possible world state
d) A critic that grades its behaviour
Q6. What is the key advantage of a utility-based agent over a goal-based agent?
a) It needs no model of the environment
b) It never needs to search or plan
c) It can rank goal-achieving outcomes by how desirable they are
d) It works only in fully observable environments
Q7. In a learning agent, which component suggests exploratory actions that may lead to new and informative experiences?
a) Critic
b) Problem generator
c) Performance element
d) Learning element
Q8. What is the role of the critic in a learning agent?
a) It chooses the external action to execute
b) It stores the transition model of the world
c) It generates new goals for the agent
d) It tells the learning element how well the agent is doing against a fixed performance standard
Q9. The test proposed by Alan Turing in 1950 judges a machine to be intelligent if:
a) a human interrogator, conversing by text, cannot reliably tell it apart from a human
b) it can prove mathematical theorems faster than humans
c) it passes a standard IQ test
d) it can learn any task from examples
Q10. The 'total Turing test' adds which capabilities to those needed for the standard Turing test?
a) Theorem proving and planning
b) Natural language processing and knowledge representation
c) Computer vision and robotics
d) Machine learning and automated reasoning
Q11. John Searle's Chinese Room thought experiment is mainly an argument against which claim?
a) That machines can ever act intelligently
b) That running the right program is by itself sufficient for genuine understanding
c) That the Turing test is too difficult to pass
d) That symbolic AI is computationally intractable
Q12. Which statement best defines a rational agent?
a) An agent that always achieves the best actual outcome
b) An agent whose actions are identical to those of a human expert
c) An agent that knows the outcome of every action in advance
d) An agent that selects the action expected to maximise its performance measure, given its percept sequence and built-in knowledge
Q13. A vision system on a conveyor belt inspects each part and decides accept or reject, with no decision affecting later parts. Which environment property does this illustrate?
a) Episodic
b) Sequential
c) Multi-agent competitive
d) Semidynamic
Q14. Which equation correctly summarises how an agent is built?
a) Agent = sensors + actuators
b) Agent = percepts + actions
c) Agent = architecture + program
d) Agent = model + utility
Q15. A table-driven agent has 4 possible percepts and a lifetime of 3 time steps. How many entries does its lookup table need to cover every possible percept sequence?
a) 12
b) 84
c) 64
d) 81
Q16. Texas hold'em poker against other players is best described as:
a) Fully observable, deterministic, single-agent
b) Fully observable, stochastic, multi-agent
c) Partially observable, deterministic, single-agent
d) Partially observable, stochastic, multi-agent
Q17. Who coined the term 'Artificial Intelligence' in the proposal for the 1956 Dartmouth workshop?
a) Alan Turing
b) Marvin Minsky
c) John McCarthy
d) Herbert Simon
Q18. Which description of the automated taxi driving environment is correct?
a) Fully observable, deterministic, discrete and static
b) Partially observable, stochastic, dynamic, continuous and multi-agent
c) Fully observable, episodic, static and single-agent
d) Partially observable, deterministic, discrete and episodic
Quick revision notes
- PEAS = Performance measure, Environment, Actuators, Sensors.
- Agent = architecture + program; the agent function maps percept sequences to actions.
- Rational = maximise expected performance given percepts and prior knowledge, not omniscience.
- Agent types in increasing generality: simple reflex, model-based reflex, goal-based, utility-based; any can be made a learning agent.
- Learning agent parts: performance element, critic, learning element, problem generator.
- Environment dimensions: observable, agents, deterministic/stochastic, episodic/sequential, static/dynamic/semidynamic, discrete/continuous, known/unknown.
- Total Turing test = Turing test + computer vision + robotics.
Frequently asked questions
What is the difference between an agent function and an agent program?
The agent function is an abstract mathematical mapping from every possible percept sequence to an action. The agent program is the concrete code that implements that function on a physical architecture with limited memory and time.
Why is a simple reflex agent not enough for most real environments?
It reacts only to the current percept, so in partially observable environments it cannot remember what it has seen and may loop or act wrongly. Adding an internal state, as in a model-based agent, fixes this.
What is a semidynamic environment?
It is an environment that does not itself change while the agent deliberates, but the agent's performance score does change with time. Chess played with a clock is the standard example.
Comments
Post a Comment