Artificial Intelligence

Question 1

What is the Turing Test?

Answer 1

behaviour-based

• computer must be able to convince person in another room that it can answer questions like a human would
• should convince people that any computer that could pass this test fair-and-square would be intelligent

DOES NOT care whether machine is intelligent or not

DOES care whether machine acts/behaves like it’s intelligent

Question 2

What is the Searle Chinese Room Test?

Answer 2

internal-state

• computer that can translate languages is NOT intelligent because it doesn’t attach meaning to words
• algorithm being followed ≠ thinking

Question 3

What is artificial general intelligence (AGI)?

Answer 3

understand or learn any intellectual task that a human being can

Question 4

What is narrow AI?

Answer 4

don’t perform outside of the single task that they are designed to perform

Question 5

What is natural language processing (NLP)?

Answer 5

automatic processing of human language (ie. by computers)

Question 6

What is the traditional NLP approach?

Answer 6

long list of rules for processing language, formulated by people and programmed into computers

Question 7

What is the modern NLP approach?

Answer 7

machines learn from text examples using artificial neural networks and similar approaches (like babies)

• statistical methods allows to compare different interpretations

Question 8

What are the 3 advantages of modern machine-learning NLP methods vs. traditional hand-coded rules?

Answer 8

• relevance: learning automatically from bodies of text means that common cases are trained more carefully
• robust: statistical algorithms can make guesses about unfamiliar and erroneous input (ie. misspelled words)
• training: simply provide more text to further train modern algorithms

Question 9

What are the traditional NLP steps?

Answer 9

1. Recognize speech (Watson skipped this – it received ASCII versions of questions)
2. Syntax analysis, or parsing
3. Semantic analysis
4. Pragmatics

Question 10

What is parsing (syntax analysis)?

Answer 10

inferring parts of speech and sentence structure, using lexicon and grammar

takes sentence and shows how words are assigned parts of speech and build up to form a sentence

Question 11

What is semantic analysis?

Answer 11

inferring meaning using syntax and semantic rules

Question 12

What are pragmatics?

Answer 12

inferring meaning from contextual information

Question 13

How is parsing the basis for computer programming?

Answer 13

• computer has to ‘understand’ programs in order to execute them
• programming languages are designed so that they can be parsed unambiguously
• grammar specifies all possible programs that can be written in a language
• designing programming languages (and their grammars) is fun and important part of CS

Question 14

Why is parsing hard?

Answer 14

can be tricky – natural languages are ambiguous

Question 15

Which type of NLP has trouble with ambiguities?

Answer 15

traditional, rule-based NLP

Question 16

Syntax vs. Semantics

Answer 16

syntax:
- knowing if word is noun or verb (syntax) tells us something about its meaning (semantics)
- describes sentence’s structure

semantic analysis:

• provide additional information – word categories, relationship between words
• adds limited meaning that can be figured out using simple rules that don’t require much context

Question 17

Pragmatics

What do most techniques do to find semantic meaning of words?

Answer 17

look for clues in surrounding text to disambiguate word meaning

Question 18

When is pragmatics especially important?

Answer 18

when sentences contain pronouns

Question 19

What is the traditional method to learning to translate between languages?

Answer 19

1. Apply NLP to each sentence in each language

2. Follow a set of rules that define how to translate

Question 20

What is the newer method to learning to translate between languages?

Answer 20

use machine learning techniques plus data on the web

Question 21

What is machine learning?

Answer 21

process of using data to help systems ‘learn’ without being explicitly programmed how to do task

Question 22

What are 3 limitations of traditional NLP?

Answer 22

• natural language is structurally ambiguous, so parsing alone cannot lead to understanding
• synonyms for words can’t be used interchangeably in every context
• natural languages have many exceptions to grammatical rules – no agreed-upon grammar for all uses of a language

Question 23

What is the brain?

Answer 23

mass of highly interconnected neurons

• dendrites receive inputs
• axons carry output
• output is based on input

Question 24

What neuron structure gives the brain many adaptive advantages?

Answer 24

many various connections among individual neurons is more important than individual neurons themselves

Question 25

What are artificial neural networks (ANNs)?

Answer 25

many interconnected artificial neurons – outputs of some feed into others

Question 26

What can simple ANNs carry out?

Question 27

What can simple ANNs carry out?

Answer 27

basic logic functions

Question 28

How can ANNs learn?

Answer 28

- with life experience, depending on particular person’s trials and errors, synaptic connections among pairs of neurons get stronger or weaker - ANN could do something similar, by gradually altering (on guided trial-and-error basis) numerical relationships among artificial neurons - wouldn’t need to be pre-programmed with fixed rules – would instead rewire itself to reflect patterns in data it absorbed

Question 29

Can ANNs learn a specific operation?

Answer 29

yes

Question 30

How do ANNs work?

Answer 30

neurons in ANN can solve one tiny function, then pass the result on to another neuron

Question 31

What is deep learning?

Answer 31

using deep neural networks neurons in ANN solve one tiny function, and pass the result on to another neuron contains many layers of neurons – if one neuron makes mistake, it does not have profound outcome on overall effect

Question 32

What are word embeddings?

Answer 32

multi-dimensional maps of language words correspond to points in multi-dimensional space - words that are "similar" should be close in this space

Question 33

What are word embeddings created by?

Answer 33

training neural networks

Question 34

What are word embeddings the key to?

Answer 34

new and improved translation software

Question 35

How does word embedding work?

Answer 35

- need huge volume of text - train neural network to associate nearby words in texts - enter word into input layer, and network should produce nearby word in output layer - if it can do this with small hidden layer, it is somehow embedding meaning of these words

Question 36

What are some ideas to improve language translation?

Answer 36

- train artificial neural networks to recognize patterns across languages - use word embeddings as input to networks results almost as good as human translators

Question 37

What are 3 types of machine learning?

Answer 37

- supervised - unsupervised - reinforcement

Question 38

What is supervised machine learning?

Answer 38

data is labelled to instruct machine which patterns to look for

Question 39

What is unsupervised machine learning?

Answer 39

data is unlabelled, machine looks for any patterns

Question 40

What is reinforcement machine learning?

Answer 40

learn by trial and error to achieve a clear objective

Question 41

What is machine learning?

Answer 41

looking for patterns in massive amounts of data

Question 42

What are adversarial attacks?

Answer 42

inputs to machine learning models that attacker has intentionally designed to cause model to make mistake – like optical illusions for machines