Introduction

Question 1

What does it mean to “act humanly?”

Answer 1

This pertains to the Turing Test approach of AI. Teaching AI to mimic human behaviour.

Question 2

What is the Turing Test?

Answer 2

A thought experiment that would sidestep philosophical vagueness of the question, “can machines think?” A computer passes the test if a human interrogator, after posing some written questions, cannot tell whether the written response comes from a person or from a computer.

Question 3

What is the Winograd Schema Challenge?

Answer 3

A test of machine intelligence designed to be an improvement on the Turing test. It is a multiple-choice test that employs questions to be answered by applying the resolution of anaphora. the machine must identify the antecedent of an ambigious pronoun in a statement.

Question 4

What is an example of the Winograd Schema Challenge?

Answer 4

“The trophy doesn’t fit in the suitcase because it’s too large.” - “it’s” refers to the trophy is this sentence.
“The trophy doesn’t fit in the suitcase because it’s too small.” - “it’s” refers to the suitcase in this sentence.

Question 5

What is the Chinese Room Argument?

Answer 5

Imagine a person who doesn’t speak chinese is standing in a room, translating Chinese to English following a set of instructions. Does the person in the room understand Chinese? It argues that syntax is not semantics.

Question 5

What does it mean to “think humanly.”

Answer 5

The automation of activities that we associate with human thinking, activities such as decision-making, problem solving, learning etc.

Question 6

What interesting findings were there from the hermaphrodite nematode study?

Answer 6

Scientists recreated the nematode brain (302 neurons, 132 muscles and 26 non-muscle end organs).

Question 7

What does it mean to think rationally?

Answer 7

The solving of problems through logic-based thinking.

Question 8

What is a syllogism?

Answer 8

A logical structure that applies deductive reasoning to arrive at a conclusion given true premises.

Question 9

What does it mean to “act rationally.”

Answer 9

This refers to an agent acting in an environment - the creation and study of intelligent agents.

Question 10

What are the three criteria for an agent acting intelligently?

Answer 10

1. Actions are appropriate for it’s goals and circumstances
2. It’s flexible to changing environments and goals
3. It learns from experience

Question 11

What are two schools of though on AI?

Answer 11

1. Weak AI - the belief that machines can be made to act as if they were intelligent.
2. Strong AI - the belief that machines are actually thinking.

Question 12

What are the two goals of AI?

Answer 12

1. Scientific goal: to understand the principles that make intelligent behaviour possible in natural or artificial systems.
2. Engineering goal: design useful, intelligent artifacts

Question 13

What is an agent?

Answer 13

An agent is an entity that perceives its environment through sensors and acts on its environment through effectors/actuators. An agent can be seen as a mapping between percept sequences and actions.

Question 14

What can be said about an agent that relies less on built-in knowledge?

Answer 14

The agent is more autonomous.

Question 15

What is a percept?

Answer 15

The input that an intelligent agent is perceiving at any given moment.

Question 16

What is a rational agent?

Answer 16

A rational agent is:
- the performance measure defining the agent’s degree of success
- the percept sequence, the sequence of all things perceived by the agent
- the agent’s knowledge of the environment
- the actions that the agent can perform
For each possible percept sequence, an ideal rational agent does whatever is possible to maximize its performance, based on the percept sequence and its built-in knowledge.

Question 17

What is the task environment?

Answer 17

PEAS - Performance measure, Environment, Actuators and Sensors

Question 18

What is the simulated vs Situated environment?

Answer 18

Simulated: a separate program is used to simulate an environment, feed precepts to agents, evaluate performance.
Situated: the agent acts directly ont he actual environment the agent has a physical body in the world.

Question 19

What is embodiment?

Answer 19

The robots have bodies and experience the world directly - their actions are part of the dynamics with the world, and actions have immediate feedback on the robot’s own sensations.

Question 19

What are static vs dynamic environments?

Answer 19

Static: The environment does not change while the agent is thinking
Dynamic: the environment may change while the agent is thinking

Question 19

What are discrete vs continuous environments?

Answer 19

Discrete: only a finite (or countable) number of discrete percepts/actions
Continuous: states, percepts, or actions can vary continuously.

Question 20

What are fully observable and partially observable environments?

Answer 20

Fully observable: agent percept contains all relevant information about the world.
Partially Observable: some relevant information is hidden from the agent.

Question 21

What is deterministic vs Stochastic environments?

Answer 21

Deterministic: the current state uniquely determines the next state.
Stochastic: there is some random element involved

Question 22

What are episodic vs Sequential environments?

Answer 22

episodic: Every action by the agent is evaluated independently Sequential: the agent is evaluated based on a long sequence of actions

Question 23

What are unknown vs known environments?

Answer 23

Unknown: the rules of the game or physics/dynamics of the environment are partially/fully unknown to the agent. Known: the rules of the game, or physics/dynamics of the environment are known to the agent.

Question 24

What is a single-agent vs multi-agent environment?

Answer 24

Single agent: there is only one agent operating in the environment Multi-agent: there are more than one agent's operating in the environment

Question 25

What is a reactive agent?

Answer 25

An agent that chooses the next action based solely on what they currently percieve, using a policy or set of rules which are simple to apply.

Question 26

What are the limitations of reactive agents?

Answer 26

Reactive agents have no memory or state, they are unable to base decisions on previous observations. They may repeat the same sequence of actions over and over.

Question 27

What is a model-based agent?

Answer 27

Model-based agents are the most effective way to handle partial observability. The agent keeps track of the current state of the world, using an internal model. It then chooses an action in the same way as the reflex agent.

Question 28

What are the limitations of a model-based agent?

Answer 28

These agents will perform poorly when the task requires searching several moves ahead, complex tasks requiring many individual steps and logical reasoning to achieve goals.

Question 29

What is a planning agent?

Answer 29

A planning agent will construct plans to achieve it's goals, and then execute them. Has a world model and a planning model which feeds into the actions.

Question 30

What is a goal-based agent?

Answer 30

An AI agent that is designed to achieve a specific goal. A planning, goal-based agent is more flexible because the knowledge that supports its decisions is represented explicitly and can be modified.

Question 30

What is a utility-based agent?

Answer 30

A model-based agent uses a model of the world, along with utility function that measures it preferences among states of the world. IT chooses the action that leads to the best expected utility, where expected utility is computed by averaging over all possible outcome states, weighted by the probability of the outcome.

Question 31

What is a game-playing agent?

Answer 31

An agent with an opponent model, world model and planning model.

Question 32

What is a learning agent?

Answer 32

Learning is not a separate module, but rather a set of techniques for improving the existing modules.