Exam: Introduction to Artificial Intelligence (AI) and Machine Learning (ML) (August 2023)

 What is the difference between AI and ML?

AI is a subset of ML

are two distinct but complementary technologies - c

ML is a subset of AI

They are two names for the same thing

To train an ML Model I need...

Lots of data and humans to label it

Lots of data and either humans or an automated way to label it

To hire a consulting company that specializes in AI

Data and labeling, the amount of data and the type of labeling depends on the use case

Which of the following is an example of AI?

Self-driving cars

Netflix

Automated Voice Assistants (Siri, Alexa, Cortana, …)

All of the above

Deep Learning is...

A ML training technique that takes a long time, because it has to "think deeply" about the task

An acronym for Data, Examples, Evaluation, Prototyping

A type of ML that uses many interconnected layers of artificial neurons

A type of ML that uses statistical regression

What is the relationship between Generative AI and Large Language Models?

Many GenAI applications are built using LLMs

No relationship, they are distinct technologies

Only GenAI that processes text use LLMs

They are two names for the same thing

When were artificial neural networks invented?

1980's

1950's

2010's

2000's

Why do these latest advancements in AI seem particularly relevant to Cybersecurity?

The ability to easily generate reasonably sounding text will cause an influx of new malicious content/lures

These are the first genuinely humanlike chatbots we've seen

Understanding intent is at the core of a person centric security strategy

A & C

What goes into building a ML model?

Data collection

Data Curation

Labeling Examples

All of the above

What are some of the cautions of using LLMs?

You can only run LLMs on CPUs

Sometimes LLM provide authoritative sounding answers that are not true

OpenAI owns the patents on the technology and you will need to pay royalties to use

They only work with English text

What is hallucination?

When a model gets caught in a loop and doesn't provide any information

When a model answers in a language different than the input

When the model provides an answer that isn't true

When a model repeats the same response regardless of input

How would you expect to see AI utilized in the next 5-10 years?

In new products or amazing features in existing products

AI usage will fade and take a similar path as Crypto and the Metaverse

Augmenting knowledge workers capabilities and accelerating their workflows

A & C

In ML terms a "parameter" most closely corresponds to what in biological analogy?

Nervous System

Synapses

Brain Stem

Nerves

How many parameters are estimated to be in GPT-4?

About 10 Million

Between 1 and 2 Trillion

Around 150 Billion

Over 500 Trillion

Which decade did we see broad emergence of Deep Learning?

1950's

1980's

2010's

2020's

LLMs _________________

can infer intent

can emit novel content

translate from one language to another

All of the above