ai lecture

An Introduction
to Generative AI

Cori Faklaris

Assistant Professor, Dept. of Software and Information Systems

Charlotte AI Institute for Smarter Learning, UNC Charlotte Dubois Center, May 18, 2023

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 2

Key takeaways from this talk

●Generative AI tools are great for PRODUCTIVITY - they can be nifty shortcuts
to dispose of low-value tasks and / or to jumpstart creativity

●Generative AI tools should always be used - and taught to be used - with a
critical mind, because they are prone to mistakes and “hallucinations”

Overview of Generative AI

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 4

Lots of hype - and doom /gloom - around AI right now …

It’s difficult to know where to look or how to start to understand AI

We tend to be afraid of things that we don’t understand

#evilbrag

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 5

When you hear “AI,” think “statistical pattern-matching”

●Oracle describes AI this way:

AI has become a catchall term for
applications that perform complex tasks
that once required human input, such as
communicating with customers online or
playing chess.

The term is often used interchangeably
with … machine learning (ML) and deep
learning.

Text from What is Artificial Intelligence (AI)? Oracle, n.d. Retrieved May 16, 2023 from https://www.oracle.com/artificial-intelligence/what-is-ai/
Image from Pattern Recognition. GeeksforGeeks. Retrieved May 16, 2023 from https://www.geeksforgeeks.org/pattern-recognition-introduction/

The data is “tokenized”(= made
into “chunks” of words, punctuation
marks, pixels, etc.) during this
process - remember this for later

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 6

AI has been with us for years, whether “generative” or not

Google and other search engines

Social media recommendations

Conversational user interfaces such as Siri and Alexa

Sensor-informed driver assistants in cars and trucks

“Auto-complete” and “smart replies” for email and text messaging

Tiktok screenshots from J. D. Biersdorfer. 2022. The Latecomer’s Guide to TikTok. The New York Times. Retrieved May 16, 2023 from https://www.nytimes.com/2022/10/26/technology/personaltech/tiktok-guide-latecomers.html
ADAS images from Wikipedia contributors. 2023. Advanced driver-assistance system. Wikipedia, The Free Encyclopedia. Retrieved from https://en.wikipedia.org/w/index.php?title=Advanced_driver-assistance_system&oldid=1150142876

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 7

Now, AI can synthesize part or all of a creative work

●McKinsey defines generative AI as:

… Algorithms (such as ChatGPT) that can

be used to create new content, including

audio, code, images, text, simulations, and

videos.

Recent breakthroughs in the field have the

potential to drastically change the way we

approach content creation.

Text and image from What is generative AI? McKinsey. Retrieved May 16, 2023 from https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 8

How Generative AI works (admittedly oversimplified)

The system generates text or images using its previously built model of the
statistical distributions of tokens(= “chunks” of words, punctuation marks,
pixels, etc.) created from its very large training dataset.

Image from Pattern Recognition. GeeksforGeeks. Retrieved May 16, 2023 from https://www.geeksforgeeks.org/pattern-recognition-introduction/
Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

Doc

Chat

Image

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 9

How Generative AI works (admittedly oversimplified)

It might make mistakes or “hallucinate” based on the limitations of its
process, but the output still might look like what you wanted.

Ted Chiang’s analogy = “unreliable photocopier” or a “blurry JPEG”

Ted Chiang. 2023. ChatGPT Is a Blurry JPEG of the Web. The New Yorker. Retrieved May 10, 2023 from https://www.newyorker.com/tech/annals-of-technology/chatgpt-is-a-blurry-jpeg-of-the-web
Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

Doc

Chat

Image

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 10

How Generative AI works (admittedly oversimplified)

Ted Chiang’s analogy = “unreliable photocopier” / blurry JPEG image

https://commons.wikimedia.org/wik
i/File:Blurry_eiffel.jpg - shared
under CC-SA 4.0 license

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 11

How Generative AI works (admittedly oversimplified)

We can ask it
questions - but a
very specific type
of question known
as prompts,
following this
structure:

Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551

“Here’s a fragment of text.

Tell me how this fragment might <continue on in
this language, or suggest a particular image>.

According to your model of the statistics of
<human language, or human-handled images>,
what <words, or pixels> are likely to come next?”

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 12

How Generative AI works (admittedly oversimplified)

The prompts are converted into tokens (= “chunks” of words, punctuation marks,
pixels, etc.), then the system analyzes what is likely to come next, based on the
tokens in its own dataset (as many as 32,000 in GPT-4!).

It then generates a tokenized output.

Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

n tokens in

1 token out

Vector of

probabilities from

own tokens

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 13

How Generative AI works (admittedly oversimplified)

With each output, it keeps re-analyzing the probabilities to decide next tokens.

Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

n tokens in

1 token out

She went to the store and

shopped

Vector of

probabilities from

own tokens

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 14

HERE’S THE REALLY COOL PART!!!

Transformers (the “T in “GPT”) know how to direct attention to specific parts
of the input to guide their selection of the output - such as verb tenses, objects.

Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention Is All You Need. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/1706.03762
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

n tokens in

1 token out

She went to the store and

shopped

Vector of

probabilities from

own tokens

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 15

How Generative AI works (admittedly oversimplified)

The system can give you different answers to the same inputs:

Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

n tokens in

1 token out

She went to the store and

bought

Vector of

probabilities from

own tokens

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 16

How Generative AI works (admittedly oversimplified)

The system can give you different answers to the same inputs:

Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

n tokens in

1 token out

She went to the store and

clocked in

Vector of

probabilities from

own tokens

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 17

How Generative AI works (admittedly oversimplified)

The system can give you different answers to the same inputs:

Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

n tokens in

1 token out

She went to the store and

danced

Vector of

probabilities from

own tokens

huh?

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 18

How Generative AI works (admittedly oversimplified)

“Hallucinations” - when the output doesn’t seem to make sense - are why it is
important not to accept everything it outputs at face value.

Murray Shanahan. 2022. Talking About Large Language Models. arXiv [cs.CL]. Retrieved from http://arxiv.org/abs/2212.03551
Bea Stollnitz. How generative language models work. Retrieved May 10, 2023 from https://bea.stollnitz.com/blog/how-gpt-works/

n tokens in

1 token out

She went to the store and

danced

Vector of

probabilities from

own tokens

huh?

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 19

Examples of publicly available Generative AI tools

Crowdsourced list of
available AI tools:
https://bit.ly/UsefulLLMs

Using Generative AI as an academic

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 21

Use DALL-E 2 to create images for course slides

-Goal 1: Quickly
source visuals
that add interest
and reinforce
content

-Goal 2:
Demonstrate
limits of AI output
with limited inputs
or prompts

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 22

Use ChatGPT to create first draft of biography text

-Goal 1: Cut the
time spent on
low-value but
necessary job
tasks

-Goal 2: Goof
around with
fellow academics
on social media

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 23

Use BingChat to draft a grant proposal

-Goal 1: Overcome
“analysis
paralysis”, make
yourself laugh in
the process

-Goal 2:
Experiment with a
sequence of
prompts for
sophisticated
outputs

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 24

Assign students to
pick/use a tool, then
critique the output

-Goal 1: Give permission
and encouragement to
play around with new
tech

-Goal 2: Mentor class
members in how to think
critically use of AI tools

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 25

My syllabus policy on “Use of AI and Other Creative Tools”

In this course, students are permitted to use tools such as Stable Diffusion, DALL-E, ChatGPT, and BingChat. In general, permitted use of such tools is consistent with permitted use of non-AI assistants such as Grammarly, templating tools such as Canva, or images or text sourced from the internet or others’ files.

No student may submit an assignment or work on an exam as their own that is entirely generated by means of an AI tool.

If students use an AI tool or other creative tool to generate, draft, create, or compose any portion of any assignment, they must (a) credit the tool, (b) identify what part of the work is from the AI tool and what is from themselves, and (c) briefly summarize why they decided to use the tool and include its output.

Cori Faklaris. 2023. Policy on Use of AI Tools for my course syllabus, version 1.0. Cori Faklaris’ blog – HeyCori. Retrieved May 16, 2023 from
https://blog.corifaklaris.com/2023/03/17/policy-on-use-of-ai-tools-for-my-course-syllabus-version-1-0/#.ZGPtWezMJqs

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 26

Some actual and/or realistic risks of using generative AI

●Violations of data privacy

○Some students told me they do not feel comfortable giving up any data to such services, such
as may be required for creating an account. For these students, I created an alternate
assignment for Slide 24, using a search engine.

●Violations of intellectual property

○Check the Terms of Service - will your inputs or prompts be used as training data?

●Violations of academic integrity

○Do a spot check of outputs, using a search engine, to see if any are wholly from another work

○Analyze submitted work using Open AI’s AI Text Classifier or the multi-service GPTZero

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 27

Humans’ #1 skill set will continue to be communication

Screenshot from
https://twitter.com/TheRealOllieLaw/status/
1656605938374307840?s=20

An Introduction to Generative AI | Cori Faklaris, Assistant Professor, Dept. of Software and Info. Systems | Page 28

Key takeaways

●Generative AI tools can be nifty
shortcuts to dispose of low-value tasks
and / or to jumpstart creativity.

●Generative AI tools should always be
used with your “thinking cap” on
because they are prone to mistakes
and “hallucinations.”

Thank you for listening!

Crowdsourced list of
available AI tools:
https://bit.ly/UsefulLLMs