Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Introduction to AI and Machine Learning > 02_what is AI? > Flashcards

02_what is AI? Flashcards

1
Q

What is intelligence?

A

legg and hutter (2007):
Intelligence measures an agent’s ability to achieve goals in a wide range of environments.

  • property that agent has as it interacts with its environment/s
  • related to the agent’s ability to succeed or profit with respect to some goal or objective
  • depends on how able the agent is to adapt in different objectives and environment
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is an intelligent agent and what concepts are related to it?

A

PEAS Model:
- Performance Measure
- Environment (world or dataset)
- Actuators (interact with E)
- Sensors

Agent has: Sensors that percept the E, and agent function that maps percept sequence to a set of actions.

Agent has a specific task. A performance measure quantifies how well the agent performs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What does PEAS stand for?

A

PEAS Model:
- Performance Measure
- Environment
- Actuators
- Sensors

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is a rational agent?

A

A rational Agent should select an action (based on what it perceives)
that is expected to maximize its performance measure
given the evidence provided by the percept sequence
and whatever built-in knowledge the agent has.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

AI can be seen as the study of what?

A

AI can be seen as the study of rational agents.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What are the 5 different models for the implementation of the agent function in the PEAS model?

A
  • simple reflex agents
  • model-based reflex agents
  • goal-based agents
  • utility-based agents
  • learning agents

(for the first 4, the agent function is given. with the last, the agent learns it by itself)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is a simple reflex agent?
with one example

A

acts directly upon current perceipts (eg thermostat)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is a model-based reflex agent?
with one example

A

keeps an internal “model of the world” to memorize the percept sequence (eg robot mapping the route it has taken)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is a goal-based agent?
with one example

A

agent has a defined goal and utilizes information to reach that goal (eg search algorithm)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is a utility-based agent?
with one example

A

agent can predict performance measure (= utility) into the future and act accordingly to find the best solution (eg chess computer)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is a learning agent?
with one example

A

agent learns from examples for different environments (eg deep learning)

programs the agent function by itself

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is a goal of artificial intelligence?

A

1956 Dartmouth Workshop:
“means to perform a task as good as or better than a human being”

–> then: write a calculator program

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

How can we distinguish AIs in a philosophical sense?

A

Weak AI vs Stong AI

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What is a Weak AI?

A

A system that merely simulates intelligence /
acts as if it were intelligent in performing its given task

–> narrow range of what it can do, but does that very well

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is a Strong AI?

A

A system that truly exhibits intelligence and can learn any intellectual task humans can learn

–> there is no strong AI (yet)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

How can we distinguish between weak and strong AI?

A

Turing Test

17
Q

What is the Goal of the Turing Test?

A

Test a machine’s ability to exhibit intelligent behavior indistinguishable from that of a human

18
Q

What is the Setup of the Turing Test?

A

Interrogator communicates with A and B via a test-only interface
and has to decide whether A or B is a machine based on how human-like their responses are

19
Q

What are the strengths of the Turing Test?

A
  • Simple
  • features open world discussion topics
  • emphasis on emotional rather than technical intelligence
20
Q

What are the weaknesses of the Turing Test?

A
  • may not distinguish between weak and strong AI
  • test only certifies human-like intelligence, but some human behavior is unintelligent / some intelligent behavior is inhuman
21
Q

AI systems are implemented as what?

A

intelligent agents

22
Q

What does AGI stand for?
(could also be GAI)

A

artificial general intelligence
(general artificial intelligence)

23
Q

What is an AGI?

A

Artificial general intelligence (AGI) is the representation of generalized human cognitive abilities in software so that, faced with an unfamiliar task, the AGI system could find a solution. The intention of an AGI system is to perform any task that a human being is capable of.

24
Q

What was the first implementation of ANN?

A

1958: The Perceptron

as a physical device mit manually tunable “weights” (changes the voltage)

25
Q

What is an ANN?

A

Artificial Neural Networks

26
Q

When was the first AI Winter?

A

1969, major limitations of ANNs revealed (cannot approximate XOR function)

27
Q

What were causes for the first AI Winter?

A
  • High expectations of AI not met
  • Lack of Computational resources
  • Limitations of ANNs become clear, the AI was unable to learn sensoric or perception skills
28
Q

When was the second AI wave?

A

1980s: success of expert systems (reflex based agents)

29
Q

What were causes for the second AI Winter?

A
  • limited applicability
  • inability to learn
  • despite the introduction of backpropagation for AIs to learn, the hardware was not ready yet
30
Q

When was the third AI wave?

A

1990-2000s: slow progress due to improved computational resources

31
Q

What is AlexNet?

A

2012:
Convolutional Neural Network trained on two GPUs with backpropagation
- beats all challengers on the ImageNet Challenge with a wide margin

32
Q

When and why did AI really take off?

A

2010s: Deep Learning
Deep Neural Networks outperform almost all other ML methods based on two factors:

  • training on GPUs is highly efficient
  • vast amounts of data is available to train and validate these models
33
Q

What are AI implementations coming up today?

A
  • Natural language processing
  • generative models
  • stable diffusion (from noise to high resolution)
  • protein folding (creating medicine)
  • games (alphaStar
  • autonomous driving
34
Q

How could stable diffusion be corrected (eg when there are too many hands in a picture)?

A

could use knowledge graph for constraints, but is very complex and not easily implemented

Introduction to AI and Machine Learning (10 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions