AI - An Introduction

Question 1

two classifications for AI

Answer 1

narrow AI
general AI

Question 2

narrow AI?

Answer 2

the ability of a system to achieve a certain goal or set of goals

majority of AI systems fall into this category

Question 3

general AI?

Answer 3

sometimes called artificial general intelligence (AGI) or strong AI

the ability to achieve unlimited goals or set new goals

Question 4

why have developments in AI progressed rapidly?

Answer 4

• growing availability of data
• growth in available computer processing power
• development of sophisticate algorithms

Question 5

generative AI?

Answer 5

branch of AI that generates content (e.g., news articles, drawings, scripts, software codes)

Question 6

difference between narrow and general ai?

Answer 6

narrow ai can do certain tasks - specific tasks

general ai can perform unlimited tasks

Question 7

one of the biggest barriers to overcome by a business looking to introduce AI is…

Answer 7

ensuring cooperation between machines and humans

Question 8

how can AI and humans work together?

Answer 8

humans can act as a backup when the limit of AI is reached

Question 9

jargon?

Answer 9

technical terminology that permits efficiency of communication

Question 10

what is AI?

Answer 10

technologies with the ability to perform tasks that would usually require human intelligence (e.g., visual perception, speech recognition, language translation)

AI has the capacity to learn or adapt to new experiences or stimuli

Question 11

what is big data?

Answer 11

high volume, high velocity and high variety information assets

for enhanced insight in decision making

data that is difficult to analyse using traditional data analysis methods

Question 12

algorithms?

Answer 12

a series of instructions for solving a problem or performing a calculation

fundamental aspect of AI systems

Question 13

processes where AI can be used in finance?

Answer 13

• robo advice
• chatbots
• fraud detection & risk management
• regulatory compliance
• stock predictions
• credit approval

Question 14

machine learning?

Answer 14

fast-growing form of AI which gives computers the ability to learn without being explicitly programmed

learns from data

uses algorithms together with supervised and unsupervised learning

Question 15

two types of machine learning?

Answer 15

supervised & unsupervised

supervised = algorithms developed based on datasets; algorithms have been trained

unsupervised = algorithms aren’t trained

Question 16

deep learning?

Answer 16

uses many layers of artificial neurons to solve more difficult problems

any artificial neural network that has more than one layer

‘deep’ refers to number of layers of thought

Question 17

what happens when data is fed into a deep neural network?

Answer 17

each artificial neuron transmits a signal to linked neurons in the next level

Question 18

differences between machine and deep learning?

Answer 18

• machine learning only works well with structured data, deep learning works well with structured & unstructured data
• machine learning can’t easily perform complex tasks, deep learning can
• machine learning needs labelled sample data to extract patterns, deep learning accepts large volumes of data and can extract features
• machine learning performance decreases as amount of data increases; deep learning amount of data has no impact on performance of algorithms

Question 18

‘human-in-the-loop’

Answer 18

where a human being is part of the feedback loop that machines learn from

occurs when machines aren’t mature enough and require human oversight

Question 19

predictive coding?

Answer 19

a type of machine learning technology to help predict how documents can be classified based on limited human input

uses supervised learning - it relies on human experts to work out and test associations

Question 20

actions for predictive coding?

Answer 20

• agree on coding protocol
• date range is decided upon
• documents that don’t have sufficient text are excluded
• repeated content is removed

Question 21

natural language processing?

Answer 21

the ability of software to read & understand a variety of documents

Question 22

robotic process automation (RPA)?

Answer 22

software programmes designed to automate tasks by mimicking humans

Question 23

rules based systems?

Answer 23

capture & used experts knowledge to provide solutions

Question 24

computer vision?

Answer 24

ability to identify objects, scenes etc in naturally occurring images

Question 25

speech recognition?

Answer 25

transcribes human speech automatically & accurately

Question 26

risks with AI

Answer 26

• algorithmic bias
• reputational risks
• legal risks & liabilities
• overestimating the capabilities of AI

Question 27

values & principles for the use of AI in business?

Answer 27

accuracy
respect of privacy
transparency
interpretability
fairness
integrity
control
impact
accountability

Question 28

AI chatbots hallucinating?

Answer 28

chatbot may generate false, misleading or nonsensical statements - chatbot is hallucinating

Question 29

AI chatbot censorship

Answer 29

chatbots in some countries are becoming increasingly censored in what they can say

Question 30

AI chatbot guardrails?

Answer 30

safety controls put in place to prevent AI from being used for harmful purposes

e.g., programming the chatbot to disable users asking unethical/illegal queries