Online Class 25 July

Summary

In this online class, the instructor discusses the responsible use of generative AI, highlighting its history, current state, and ethical concerns such as author compensation and environmental impact. They address AI's limitations, particularly its bias towards English-speaking contexts, and the importance of human intervention to provide localized insights. Students are encouraged to utilize AI for language learning and to create outlines for assignments, while also being aware of AI's generalizations and the need for personal experience to enrich discussions.

Takeaways

• 📚 The instructor is attending a conference sponsored by Conan University and will present research for feedback, highlighting the importance of academic engagement.
• 🗓 The class schedule includes an online course this week and another in two weeks, with a reminder for students to meet on August 1st.
• ⏰ Students are advised to manage their time effectively, spending no more than one hour on lectures and exercises provided through the Learning Management System (LMS).
• 🎓 The assignment involves creating a first draft of a presentation, emphasizing the importance of demonstrating understanding of class concepts, signal phrases, verb forms, and rhetorical patterns.
• 📝 There is no need for a fancy video; the focus should be on content awareness and proper use of English language tools.
• 🔗 The LMS provides resources such as a worksheet and a platform for uploading files, with a suggestion to work during normal class time for structure.
• 💬 The instructor is available for questions and comments via email or Facebook, indicating a supportive and accessible teaching approach.
• 🤖 The lecture discusses generative AI, its history, and ethical considerations, including author compensation and environmental impact.
• 🌐 Generative AI's limitations are highlighted, particularly its bias towards English-speaking countries and the lack of personal experience in its responses.
• 📈 The controversy surrounding Taylor and Francis selling texts to Microsoft for AI training underscores the ongoing debate about data use and consent in AI development.
• 🚫 The potential for AI to facilitate cheating and misinformation is acknowledged, along with the need for responsible and ethical AI usage in academic and professional settings.
• 🌟 The opportunity for non-English speakers to contribute unique perspectives to AI is presented as a positive aspect of the current state of generative AI.

Q & A

What is the main purpose of the online class mentioned in the transcript?

The main purpose of the online class is to provide an opportunity for the instructor to present their research, receive feedback, and guide students through the course material, including lectures and exercises.

What is the instructor's upcoming event at Conan University?

The instructor is attending a conference sponsored by Conan University, where they will present their research the following afternoon.

How long is the instructor's presentation expected to take?

The presentation should take no more than one hour, with an initial expectation of 30 minutes.

What is the assignment for the students during the online class?

The assignment involves listening to the lecture, completing an exercise using the links on the Learning Management System (LMS), and working on the first draft of their presentation.

What is the instructor's advice regarding the presentation's format?

The instructor advises that there is no need for a fancy video; the most important aspect is to demonstrate awareness of class concepts, including signal phrases, verb forms, and rhetorical patterns.

What is the controversy mentioned in the transcript about Taylor and Francis?

The controversy is that Taylor and Francis, a major academic publisher, sold all of its texts to Microsoft to be used in training a large language model (LLM), without informing or consulting the authors, raising ethical questions about compensation and consent.

What are some of the ethical concerns associated with generative AI?

Ethical concerns include compensation for authors whose work is used to train AI models, environmental impact due to data center energy consumption, potential for cheating in educational and professional settings, and the spread of fake news and images.

What is the bias issue with large language models mentioned in the transcript?

The bias issue is that large language models are predominantly trained on data from English-speaking countries, which can lead to limited understanding and inappropriate suggestions when dealing with non-English contexts or cultures.

How can generative AI be used responsibly in the context of the class?

Generative AI can be used responsibly by asking for advice on language use, requesting outlines for speeches or assignments, and critically analyzing the AI's responses for cultural and contextual relevance.

What is the instructor's suggestion for students to engage with generative AI?

The instructor suggests that students use a free version of generative AI like chat GPT to ask questions about language use and outlines for specific topics, and then analyze the responses in the context of their assignments.

How can students submit their questions and analysis to the instructor?

Students can submit their questions and analysis through a Google form provided on the LMS, where the instructor will review them and provide feedback.

Outlines

00:00

📚 Online Class Introduction and Generative AI Discussion

The speaker welcomes the audience to an online class and mentions an upcoming conference at Conan University where they will present their research. They appreciate the participants' involvement and provide a brief overview of the course progress. The speaker instructs the class to listen to a lecture and complete an exercise within 30 minutes to one hour, emphasizing the importance of time management. They also remind the students to work on a presentation draft, focusing on class concepts, signal phrases, verb forms, and rhetorical patterns. The speaker discusses generative AI, its history, and its evolution, highlighting the recent controversies surrounding the use of academic texts by major publishers like Taylor and Francis for AI training. Ethical concerns about compensation for authors and environmental impact due to data center usage are also addressed.

05:00

🌐 Ethical Concerns and Bias in Generative AI

The speaker delves into the ethical issues surrounding generative AI, focusing on author compensation and environmental concerns due to data center operations. They discuss the potential for AI to facilitate cheating in educational settings and the creation of fake news and images. The speaker also highlights the intrinsic bias in large language models, which are predominantly influenced by English-speaking countries. This bias can lead to inaccurate or irrelevant responses when discussing non-English contexts or regions outside the US, UK, Australia, and Canada. The speaker uses examples from their students' work to illustrate the limitations of AI in understanding local cultures and contexts, such as the potential impact of Amazon Go in Thailand.

10:02

🤖 Limitations of AI in Understanding Local Contexts

The speaker continues to discuss the limitations of AI, particularly in understanding local contexts and cultures. They use examples from their students' work to show how AI-generated responses can be misleading or irrelevant when applied to specific situations, such as the impact of Amazon Go in Hat Yai. The speaker points out that AI lacks personal experience and cannot provide nuanced insights into local cultures or the specific needs of non-English-speaking countries. They also discuss the potential for AI to replace human jobs and the importance of considering the broader implications of AI technology, such as its impact on local businesses and the environment.

15:03

📝 Suggestions for Using Generative AI in Language Learning

The speaker provides suggestions on how to effectively use generative AI, particularly for language learners. They recommend using AI for advice on writing and speaking, as well as for generating outlines for speeches and assignments. The speaker emphasizes the importance of personal experience and context in providing meaningful contributions to discussions, which AI lacks. They also highlight the potential for AI to assist in language learning by providing examples and suggestions for writing about other people's ideas. The speaker encourages students to use AI as a tool for generating ideas and starting points for their own work, while also considering the limitations and biases inherent in AI-generated content.

20:06

📚 Assignment Instructions and Encouragement to Engage with AI

The speaker concludes the lecture by outlining the assignment for the day, which involves using AI to ask two questions: one related to language skills and another about a specific topic for an outline. They encourage students to use free versions of AI services and to submit their questions and analyses through a Google form on the LMS for feedback. The speaker also reminds students to complete the form within the next 10 to 20 minutes and to proceed with their presentation and recording. They end the lecture by thanking the audience and inviting any questions, promising to see them in the next session.

Keywords

💡Online Class

An online class refers to a learning session that takes place over the internet, allowing students to participate remotely. In the video script, the speaker mentions being in 'non Kai' for a conference and conducting an online class, indicating the flexibility and accessibility of educational platforms in the modern era.

💡Research Presentation

A research presentation is a formal session where an individual shares their findings or work with an audience, often seeking feedback. The speaker in the script is preparing to present their research at a conference sponsored by Conan University, highlighting the importance of academic discourse and peer review.

💡LMS (Learning Management System)

LMS stands for Learning Management System, a software application for the administration, documentation, tracking, reporting, and delivery of educational courses, training programs, or learning and development programs. The script mentions using LMS links for accessing lectures and exercises, demonstrating its role in facilitating educational content delivery.

💡Generative AI

Generative AI is a type of artificial intelligence that focuses on creating new content, such as text or images, based on statistical probabilities of concepts and words. The script discusses the history and current state of generative AI, including its applications in services like Google Autocorrect and the ethical concerns surrounding its use.

💡Large Language Model (LLM)

A Large Language Model (LLM) is a machine learning model that is trained on a large dataset of text and can generate human-like text based on the input it receives. The script explains that LLMs are used in generative AI to determine what words and concepts are likely to occur together, forming the basis for AI-generated content.

💡Ethical Concerns

Ethical concerns in the context of the video script refer to the moral implications and potential negative impacts of using generative AI, such as compensation for authors, environmental impact, and the risk of spreading misinformation. The speaker discusses these concerns to highlight the responsibilities associated with AI development and use.

💡Bias in AI

Bias in AI refers to the inherent prejudices or favoritism shown by AI systems, often due to the data they are trained on. The script points out that AI trained on predominantly English-speaking internet data may not accurately represent or understand non-English contexts, leading to biased outputs and limited perspectives.

💡Amazon Go

Amazon Go is a technology-powered, cashier-less retail store concept by Amazon. The script uses Amazon Go as an example to discuss the potential pros and cons of implementing such technology in different cultural contexts, like Thailand, and how AI-generated responses may not fully consider local nuances.

💡Technology Ban

A technology ban, as mentioned in the script, refers to the prohibition of certain technologies, particularly in educational settings. The speaker uses this concept to explore the potential benefits and drawbacks of such a ban in high schools, emphasizing the need for balance between technology use and traditional learning methods.

💡Personal Experience

Personal experience in the script refers to the unique insights and perspectives individuals have based on their own lives and backgrounds. The speaker contrasts the limitations of AI, which lacks personal experience, with the value of human input in understanding and responding to complex issues, such as the impact of technology bans or the introduction of Amazon Go.

💡Language Learners

Language learners are individuals who are acquiring a new language. The script emphasizes the importance of technology for language learners, suggesting that AI can assist in language learning by providing frameworks and examples, but also noting the limitations of AI in understanding the diverse contexts in which language is learned and used.

Highlights

Introduction to the online class and the instructor's upcoming conference presentation at Conan University.

The importance of providing feedback on research during the conference for improvement.

Course progress update, mentioning the online course schedule and the instructor's availability.

Instructions for the week's assignment, emphasizing the time limit for lecture listening and exercise completion.

Clarification on the simplicity required for the presentation, focusing on class concepts and rhetorical patterns.

Details on where to find the worksheet and how to upload the presentation file on the Learning Management System (LMS).

Recommendation to work during normal class time and the option to choose a comfortable space for the task.

Invitation for students to reach out with questions or comments via email or Facebook.

Discussion on generative AI, its definition, and historical context starting with Claude Shannon's work.

Evolution of AI from early computer chess programs to modern applications like Google and Auto-correct.

The rise of generative AI in 2022 with the popularization of chatbots like Chat GPT and its competitors.

Explanation of how generative AI works, using large language models and statistical probabilities.

Controversy over Taylor and Francis selling texts to Microsoft for large language model enhancement.

Ethical concerns of generative AI, including author compensation and environmental impact.

The issue of cheating in educational institutions and the workplace due to the use of AI.

The problem of fake news and realistic-looking images generated by AI.

Addressing the bias in large language models due to the predominance of English-speaking data sources.

Example of AI-generated responses regarding Amazon Go in Thailand, highlighting the limitations of AI understanding.

The need for human intervention to provide context and personal experience in AI-generated content.

Suggestions on how to effectively use generative AI for language learning and outline creation.

Assignment instructions to use AI chat services like Chat GPT for specific inquiries and to provide feedback.

Transcripts

hello everybody and welcome to our

online class right now I'm in non Kai uh

for a conference sponsored by Conan

University tomorrow afternoon I will

present my research and it's a nice

opportunity for me to get some feedback

on my work so I appreciate you uh

participating in the online class so I

can be here uh as you know um we're

coming up about one quarter through the

course so far uh this week we have a

online course

and we'll also have an online course in

two weeks but I will see you on August

1st uh just a reminder for the plan uh

today you should listen to the lecture

and complete the exercise um using the

links on LMS that should take you 30

minutes um no more than one hour to

complete um make sure you watch your

time uh you don't want to spend too much

time on the first part because then the

second part is uh working on a

presentation um sort of the first draft

of your presentation that I'll ask you

to write and record with the rest of the

time for class I'll remind you that

there's no need for a fancy

video and the most important thing is

that you show your awareness of some of

the class Concepts the videos we analyze

together the signal phrases and the verb

forms and the rhetorical patterns um so

there also is a link on LMS for the work

worksheet and um a file to or a place

for you to upload your file um it's best

I think for you to meet during normal

class time uh but you don't have to go

to the classroom so feel free to find a

comfortable

space uh if you have any questions or

comments U please feel free to email me

or hit me up on Facebook I'll be

checking messages tomorrow uh when I

have

time so just a couple of comments about

generative AI

um as part of this class I'd like to

think about the ways in which we can

responsibly use online sources so um

thinking about the history of generative

AI uh first of all let's start out with

a definition uh generative AI is a kind

of artificial intelligence that focuses

on creating something new like text or

images the idea for AI is actually quite

old uh Claude Shannon who was a

contemporary of Alan Turing wrote some

influential papers about AI after World

War II um if you know something about

the history you know one of the first

uses and the first appearances of AI

which computer chess programs that were

able to defeat some of the world's chess

uh players um and after that um and

guess more recently you can see AI when

you're using Google or you're using Auto

correct um when you're typing on your

phone those kinds of services also are

powered by the same kind of machine

learning um that we have but things

changed a lot uh in 2022 last year uh we

started to have um generative AI that

was um popularized by chat GPT and now

there are many uh competitors um what is

how does generative AI work well briefly

stated it takes um ideas is about what's

likely to occur together what words and

concept concepts are statistically

likely to work together um programmers

get these statistics by gathering

written material into what is called a

large language model or an

llm um the initial data came from the

worldwide web and social media um but

we've sort of reached the limits of that

information and in order to expand

programmers are finding new information

to add to their llms uh last week there

was a little bit of a controversy a

Taylor and francers which is a major

academic Prof publisher bought um or

sold all of its texts to Microsoft so

that it could be used to um uh uh add to

the large language model people who have

published uh papers in Taylor and

Francis journals or published books with

the Taylor and Fran's Publishing Company

routage all of a sudden I found out that

their work is being used to train uh the

IIA the AI through the addition to the

language large language model which is

shocking to people that they were um not

told and they were not consulted on this

uh there are many uh ethical questions

about uh generative Ai and you're

probably familiar with some of them one

as you can tell this idea of

compensation for authors um people are

making money off of generative AI are

the

um are the authors being compensated for

it in most cases no uh the authors who

were publishing with Taylor and Francis

uh did not um know that there work was

going to be used and they didn't receive

any payment for those um articles uh

also um another different kind of

concern is the data center harm to the

environment uh generative AI is very

intensive use of computer processing

power which creates a lot of heat and

that means that there is um damage to

the environment every time somebody uses

uh generative AI uh of course you're

probably familiar with this uh fear that

teachers have uh about students cheating

in high school or college and also of

course uh people in the workplace who

are not doing their own work um but

using AI to um get their work done

easier

and in today's media environment fake

news and images that seem real are

startling uh today you can see pictures

of uh Donald Trump saying all kinds of

crazy things uh he certainly does say

some things that are unusual in real

life but you can also uh see videos that

have been made to look like him even

though he hasn't said those uh

words but I'd like to address one

ethical concern uh that you may not be

familiar with may not have thought about

before which is the bias that is

intrinsic to um L language learning or

large language models right where does

this data come from it comes from the

internet uh what is the internet data

like it's

predominantly speakers of English who

have written things about

English-speaking countries and so there

is a problem using the large language

model when you are talking about places

outside of the US the UK Australia

Canada those large um user bases have

contributed lots of writing that help to

tell the AI what's likely to come

together but there is a problem if you

writing in a language or speaking in a

language that is not English and you're

talking about areas that are not

familiar to the English language

speakers and this is a a problem I guess

for AI of course it's also an

opportunity uh somebody who is um

working in the United States uh might

find their job replaced by AI um but if

you want information about Thailand well

the AI doesn't seem to know too much so

let's take a look at this uh quickly uh

and look think about this bias uh that

we're

seeing when I was working with my

students in 004 last term um we were

playing around with AI a little

bit and we had that unit about Amazon go

and so I said to to them let's take a

look at what chat GPT says when we talk

about this issue of whether or not we

should bring Amazon go to

had and you can try this on your own I

use deep AI for this uh table that you

see before you

here are some of the answers I've edited

them slightly um pro version of it is

the convenience customers do not have to

wait

online uh increased access to Essentials

you can get um materials food and and

essential products uh if you don't have

a lot of stores around uh and also the

technology can help improve inventory

the cons some of the cons that um DPI

came up with are

um could potentially displace jobs uh

for existing convenience stores uh there

could be a harm to the local

businesses and the harm to Thai culture

AI claims that Tha culture places a

strong emphasis on social interactions

and Community bonding um the store where

you don't talk to the customers might be

impersonal or unfriendly by some of the

local people in Thailand so this is what

chat GPT s came up with or D came up

with um if you want to stop the video

for a second what do you think uh about

these answers in your notebook uh or if

you're sitting with somebody talk about

what you think about these answers and

then when you're ready start the video

again so this is my thinking I I'm I'm

curious to know what you think as well

um but what do you think about these

answers from chat GP

well of course one thing that students

like is that the English is very smooth

and when you read um these prompts you

say oh I wish uh my English was so clean

uh and that is definitely something true

the persuasive value of Chad GPT is

partly because of its good use of

English um but we see two types of

problems at least one of them is the

limitations of the L language model um

when we're talking about probabilities

of what words and concepts are likely to

come together those probabilities are

based on English speakers in english-

speaking countries for the most part and

so you can see that the AI does not seem

to know anything about Hadi it just

knows that it's a city uh and therefore

it makes this strange suggestion that

traffic will be um won't be so um uh

harmed uh and this idea that you know

the major concern for people in had ey

is avoiding traffic jams or causing

traffic jams is a strange idea um and

also this idea that the most important

concern is TI culture valuing social

interactions I suppose it's true that

people go to

7-Eleven with their friends sometimes

and talk but that kind of social

interaction is different from this idea

of the small store so some people you

know like to go to small stores and talk

to the owner and have a relationship uh

and that's certainly is true of places

in Thailand and others other countries

as well um but it also seems sort of

limited uh in the idea that this is uh a

big problem from Hai um also the

limitation here is that it doesn't have

any personal experience of hot yai

people who live in hot Yi go shopping at

Big stores like shopy or um lasada or so

on I mean people who want to buy things

don't necessarily go to the store to buy

them and so this um idea that people

have to go to the store is sort of

limited obviously also people haven't

been to lootus or tops or macro I mean

the idea that um people in Hadi don't

have the advantages of large stores or

shopping experiences is a little bit

strange as well uh and then of course

there's this idea that Amazon go is in

trouble if you read a little bit about

Amazon go you'll find out that Amazon is

sort of scaling back back its experience

because it's too much um too much labor

intensive for their um for their profit

margin and so something that you know we

can consider in Hai is if we invest in

Amazon ghost store I mean it could be

out of business in two years and that

doesn't come out up in the AI

suggestions at all so human intervention

here is is quite necessary let's take

another uh example from

004 we asked um chat GPT what are the

pros and cons of a technology ban in

high schools uh and the things you might

you read are sort of what you might

expect you can have better focus and

concentration enhance social skills

reduce cyber bullying and the cons are

you know maybe we should uh encourage

students to stay connected with global

networks uh if we ban technology they

won't have digital

literacy uh and the potential for

students to be cut off from each other

and this potential for isolation so

again I'd ask you to take a minute in

your notebook and discuss with your

friends if you're sitting with somebody

uh what do you think about these uh

answers um from chat GPT so you can

pause the video for a second uh and then

start again when you're

ready so here's my thinking about these

answers I mean again we see some nice

English which I think is really helpful

and friendly and and uh makes people

think that chat GPT is a good service

but when you start looking at the

answers more carefully you realize that

these answers have nothing to do with

Thailand at all you could change the

country and say what's the uh value of a

technology ban in France or what's the

the potential for a technology ban in

Nigeria and the answers wouldn't change

at all uh even the suggestion about

Global Communication doesn't really

depend on Thailand if you wanted to

think about how Thailand is you know

increasing its role in the world market

and um thinking about International

Communication as being a part of

Thailand's um new position or growing

position as a trade center and a

Transportation uh

Nexus then all of a sudden the answer

becomes real but for Chachi BT it's just

sort of this vague idea of Global

Communication of all people in the world

so again we see your personal experience

can make a big difference for your um uh

for your answers um something that chat

gbt can't really do a personal

experience also here we don't see

anything about what it's like to be a

student in a Thai High School uh so the

llm doesn't really know anything about

what it's like to be uh a student in a

Thai High School I suppose most of that

information that you find on the web is

In tha so you won't really find anything

about that um in the English language

which limits the the language learning

model um we also don't have any personal

ideas of an actual human going to high

school if you were writing about a

technology band you might talk about

your own uh personal experience of using

technology uh or how technology was

harmful while you were in school but we

don't see any of that because chat GPT

doesn't isn't a person right it doesn't

have any personal experiences

and another thing in context of what

we've been talking about in a class for

the last few weeks you know it doesn't

think about what it's like to be a a

learn a person learning English somebody

learning English can do a lot with

technology and depends on using

technology and access to English

language sources uh and so this idea

that language Learners need technology

it's not something that's on the mind of

people in English-speaking countries and

for that reason people don't write about

it on the internet and then people don't

um upload that to the large language

models and it doesn't become part of the

answers for chat GPT so again we see a

real person especially outside of an

English-speaking country has an

opportunity to add something really

great to the

conversation so that being said what can

generative AI do well here are my

suggestions for you on how you might

want to use it one thing is really great

for uh language Learners you can always

ask chat GPT for advice on what you're

trying to do for instance today you're

going to be writing a speech about

signal phrases for somebody else's ideas

well you can ask chat GPT can you give

me some suggestions on how to do that

and it'll come out real quick here's a

bunch of ways that you can uh describe

uh something somebody else's ideas and I

think that kind of um use of chat GPT or

generative AI can be really helpful uh

gives you um some ideas of what to do in

your own writing or your own speaking uh

to get ready for a

presentation um the other thing might be

um might be helpful as well oh sorry

also just not for nothing but look at

the verb tenses and the suggestions from

generative AI author's name believes

that it is argued by it is argued by

author's name suggests remember we had a

problem with that last week um but it is

true that when you're talking about

somebody else's ideas you should use the

present tense and Chad GPT knows that

and uses the present tense in those um

examples okay uh anyway second thing

that you might want to try is um um

asking for a um

uh asking for a framework an outline

right now you can ask chat GPT to get

you started on an outline this

assignment and all the assignments that

you do for class and and even when you

go into the workplace they're going to

not fit exactly in with with with what

CH chat GPT thinks right but you can ask

uh for some advice on how to get started

so if you ask an outline ask for an

outline for making a speech about the

liberal arts um you'll get a bunch of

things right I just give you the first

three points here but uh start with a

thought-provoking question introduce the

topic right talk about the definition of

the liberal arts the value of a liberal

arts education right and those kind of

things can be improved by your own

writing you can also ask a more specific

specific question uh you know you can my

question is sort of General a speech

about the liberal arts but you can ask

something with more details and say what

you're exactly you're looking for and

you might get an outline that you can

then fill in with your own uh thinking

um so again it's very general right it's

sort of you know sounds like anybody

could have written it um certainly it

applies the most to people in

English-speaking countries and this is

the problem we've been talking about in

class there's a limitation to these

ideas which is that they're focused on

people on english- speaking countries

but again I will say that's a problem

but also it's an opportunity right the

world definitely needs to hear from you

and to hear from what it's like to be a

language learner or a liberal arts

student uh and a country that doesn't

speak English as the first language and

that's a really exciting opportunity I

think for all of us

okay so your part of your assignment for

today is to give it a try um use your

favorite AI chat GPT will work um

whatever you want to use you can but

don't pay for anything uh you can use a

free version uh like um

https chat gp.com or whatever one that

you want to use uh and ask two questions

like I on like the ones in this lecture

ask one advice for langu anguage use

skills type question and ask another

question about an outline but use a more

specific topic which will give you a

better response uh and then you'll see a

link in uh on the LMS for a Google form

where you can put your um questions and

your um analysis of them into the form

so that I can read them and make some

feedback later so this first part of the

class was just about 20 minutes so

within the next um 10 to uh 20 minutes

please finish the form on uh the LMS and

then move on to your um presentation and

make that recording um for me to look at

later uh that information is also

available on the

LMS okay thank you for your attention if

you have any questions please reach out

and I will see you next week