GenAI 101 Course

Description of the video:

[Scott Shackelford:] Looking for fascinating conversations with industry thought leaders?

Want to get the inside scoop from local legends?

Interested in the stories behind the stories from celebrated visionaries?

This is your home for the best our university has to offer.

Welcome to We Are IU.

[ Music ]

Welcome to We Are IU, the podcast that brings you the voices, ideas,

and innovation shaping Indiana University and our community around the world.

I'm Scott Shackelford, Associate Vice President and Vice Chancellor for Research here at IU.

And today, we're diving into a topic that's reshaping the way we live, work,

and even imagine the future: Generative AI, and what IU is doing

to help navigate this new tech frontier.

You've probably heard about tools like ChatGPT or Gemini, but what do they really mean

for education, business, creativity, and daily life?

To help us break it down, I'm joined by Dr. Brian Williams,

Chair of the Virtual Advanced Business Technologies Department,

and an accounting professor at the Kelly School of Business; and Dr. Anne Leftwich,

the Barbara B. Jacobs Chair in Education and Technology in IU School of Education.

Dr. Williams developed IU's first practical AI curriculum for business

and helped launch Kelly's internal AI systems.

Dr. Leftwich is a nationally recognized scholar in educational technology and a strong advocate

for accessible tech-enhanced learning.

So let's get started with a couple of questions for you both, and welcome to the podcast.

[Brian Williams:] Thanks, Scott.

[Anne Ottenbreit-Leftwich:] Excited to be here.

[Scott Shackelford:] So first off, let's dig into what the heck generative AI really is.

For someone brand new to this space, what exactly -- what is generative AI?

How would you define it, and why has it exploded into public awareness so quickly?

[Brian Williams:] Yeah.

So generative AI, it's generative.

So it's generating something new, and that can be really cool.

It can be text.

It can be video.

It can be images.

And I think why it exploded so quickly is because it's so easy to work with, right?

So in the past, when I was doing my own research in AI, you have to learn a lot of math,

you have to learn a lot of programming languages like Python or things like that

to use those tools and those models.

But generative AI is totally different, right?

As long as you can write, you can interact that way, and it's really cool

and opens up a lot of possibilities.

[Scott Shackelford:] Awesome.

And why now, right?

Why has it exploded in just the last few years?

Was there a breakthrough?

[Brian Williams:] In my opinion, the technology has not been out that long.

So a lot of this is enabled through something called Transformers architecture,

and that research paper came out less than a decade ago.

And so kind of the fundamental underpinnings of this technology,

where you can generate new things from text.

It's pretty new.

And that's pretty cool.

[Scott Shackelford:] And IU is launching a Gen AI course, been very successful already,

and I know it's expanding quickly.

Why is it so important for people to understand generative AI at this moment in time?

And what drew you both to design a course like this for the IU community?

[Anne Ottenbreit-Leftwich:] I think that in terms of literacy, just as a digital literacy,

this has really transformed basically over the past decade, like Brian has said,

and really the uptick has been in the past few years, that has been increasingly more

and more important, and that AI literacy is really just becoming foundational.

And students and faculty who understand how to use it, how to critique it,

they're going to be better prepared to learn, to teach, to work in every field.

This is part of what's in the course, and we've got a few really great testimonials too

from the business side that really helped capitalize this

and talk about the importance of it.

And it's even reshaping every single discipline, too, right?

From healthcare to business to the arts.

It's really redefining professional practices, and our responsibility is

to prepare the learners, not just to use the tools,

but to shape and govern their use ethically.

[Scott Shackelford:] Yeah.

Just part of that foundational skill set, you could argue that every college student needs

to be equipped with to succeed these days.

So let's dig a little bit more into, as you were alluding to there, and AI in action.

Can you share, either you or Brian, a few examples of how generative AI is being used

in everyday life, maybe even in your own lives, at work, at school, even at home?

[Brian Williams:] Yeah.

It's being used in all sorts of ways.

So home is always fun, right?

So in our house, we have three little kids, and so we're using Gen AI to do things

like develop lesson plans for them, make learning more fun.

I use it to plan some, you know, fun date nights with my wife and try to think

of good questions to ask each other.

We're married, you know, more than a decade, so it's good to have those.

That's been really nice.

And then in the workplace, it's being used in normal ways,

like writing better emails, things like that.

But there's also some really, really cool, more advanced use cases.

So Pete Yonkman is the CEO of Cook Medical Group.

He was talking about how they're using it to develop new lifesaving technologies, right?

And even be with AI, it can also be things like generative AI, where some things I've seen

in the healthcare space are they're analyzing, you know, lots of transcripts

of how people responded to different drugs.

And you can find patterns in there that maybe a human couldn't find

to find new uses for old drugs.

And things like that, I think, are really cool and kind

of can show the potential of this technology.

[Anne Ottenbreit-Leftwich:] Absolutely.

And on the home side, similar to Brian, I have three kids as well.

And so using it for, for example, my son was trying to do a training

for cross-country to amp up his progress.

So we were able to really question, "What does this look like?"

"How do we do that?"

We like it for iterative story building.

And so this is a fun thing where we throw different pieces out there.

I've also done it for taking pictures of things that are in my fridge to figure out,

for picky eaters, what might they actually eat?

Or what might this look like?

Even ranging into taking a picture of the front of my house for landscaping purposes,

to say, "What could I fill in here?

How would this look?"

And my children have also suggested that I need to do it for taking pictures of my outfits

to see if they're actually stylish or not.

But then, on like a more personal side -- or a more professional side,

I think my usage just increases on a daily basis.

I don't know if that's the same with Brian, too.

I legitimately have really great conversations with AI brainstorming.

What are some out-of-the-box ways to think about this, or even think about ways to market it

or put it into Gen Z language so that we're reaching out to students

and really reaching them on their levels?

Using it for summarizing meetings, some of the more laborious tasks that we're forced to kind

of grapple with that just don't utilize a lot of our big brains, as my kids would say.

And then from the teaching side of things, there's so many amazing ways

for us to be able to utilize this.

We're looking at ways to personalize education.

So in the K-12 space, let's say that you have a student

in your fifth-grade classroom who's maybe reading at a first or second-grade level.

You can take any of those materials and put it in there and say, "Put this down to a first

or second grade level," even utilizing some of their key target words

that they're trying to learn how to read.

And it will take the content and information and put it into a format

that that student can actually utilize.

Or even thinking about how do I make this more relevant for my students?

Like if I'm talking about economics, I can put it into a sports analogy format

that maybe my students will be more interested to hear about it in a fantasy football type

of style to understand supply and demand and how that might look differently.

Or even thinking about, at the higher education level, how we're using it

for more along the lines of potentially grading or giving feedback to students or studying.

There's just so many possibilities.

But the point is, especially in the higher education space,

we have this really great opportunity because we are all such provocative and innovative thinkers

that we can really think about where and when AI should and could be applied.

And so I think that's our specific gift back to society -- for lack of a better term --

of really thinking through some of those pieces of what can and should it look like.

[Scott Shackelford:] I think that was a, you know, excellent survey of some of the --

some of the ways in which this technology can positively make a difference.

I remember, we used it in our own family.

We also have three kids, look at us, to do campfire songs not too long ago.

And we still sing them.

I mean, it's amazing.

Like you were saying, and the elements you can bring together in these stories,

and somehow the technology can make it work.

On the flip side, though, I'm wondering if you could both reflect a little bit on some

of the risks or misconceptions about AI that you think deserve a bit more attention.

[Brian Williams:] Yeah.

Part of AI literacy is, like, not only knowing how to use it well, right,

but also knowing what it can do poorly.

I think there's a couple here.

One big risk, especially if you're asking for advice, is it can be way too likely

to agree with whatever you say, right?

Maybe give you false confidence that maybe you --

maybe, honestly, your idea is really bad, but AI won't tell you it's bad.

It'll, you know, give you false encouragement, and that can be a real problem.

Another risk or misconception is just that it's only as good as the data it's trained on, right?

And so if you're -- if you're asking it things, you'll get a right answer a lot

of times, but sometimes you won't.

And unless you're really an expert in that area, you could be misled because it can make things

that sound beautiful and sound very convincing, but are just flat wrong, right?

So you have to be a little careful there

of going too far outside your own expertise when using it.

[Anne Ottenbreit-Leftwich:] And I think my favorite way that Brian usually talks

about it is think about it as, like, a really eager intern that just wants to please you

and doesn't necessarily have all of the right answers.

The other interesting part to this question, so when my daughters were

in fifth grade last year -- I have twin daughters,

and they actually developed their own chatbot in order to answer fifth-grade questions

because parents can't possibly understand what they're going through at this age.

And the answers that I give are always not cool, and I just don't understand.

So they designed a chatbot in order to answer questions that fifth graders might have in order

to really think about what kind of answers they could get to help them work through some

of these things that they're trying to solve around texting drama with friends

or what should they do in a situation if, you know, someone reaches out to them

and they don't know who the person is.

They don't necessarily want to have those discussions with their parents.

So they designed this chatbot and also used their own vernacular.

So it was very much in Gen Z language and used words like "slay"

and "Ohio" and other sorts of terms.

And so one of the things that they noticed, though, they started asking it questions

and it wasn't giving them the responses that they thought were appropriate, right?

So they asked about dating and then said, "Whoa, whoa, whoa, like we should not be allowed

to date at the fifth grade level," or sharing water bottles.

Like, that's not okay to share water bottles.

There's all sorts of like germs and things that can spread.

So they went back and recoded the instructions.

However, then they came up with a really interesting point of, well,

but what happens if somebody uses it

and they don't necessarily align with our value set, right?

Like what happens if their parents say it is okay to date

or it is okay to share water bottles, right?

Our code and our values are being put into that system.

So it brings up other really great points about what is the AI even trained on.

What's okay?

What do we grapple with?

What does that look like?

And I think right now, isn't it true, Brian, that many of the ChatGPT is heavily trained

on Reddit posts, which, for good or for bad, thinking about, you know, is that what we want

to further in society, or do we want to contemplate

and think about what that looks like?

And then I do know one of the other pieces is people taking it and using it

for a social crutch, which, of course, there are some really careful considerations

and concerns about who's liable for that.

What technology company -- what are technology companies doing?

And at least for me, at least in the K-12 space, it really does come down to education,

parenting, all of those things as well, to really --

you need to know what your kids are watching on TV, what they're doing,

but that's more of a K-12 answer.

[Scott Shackelford:] Yeah.

Honestly, and that is a topic I wanted to raise as well,

that you just did a great job highlighting, both of you, is just, you know,

the technology is advancing so much faster than the governance regimes, you know, for it, right?

Either, especially at the federal level, but even at the state level,

a lot of this is still being left to companies and codes of conduct.

And as a result, as you alluded to, and we're seeing some, you know,

frankly, horrific incidents, right?

So I'm thinking about, you know, the ethics involved here, you know,

some of the bias of these tools depending on the data that it's trained on, of course.

There's overarching other issues like copyright, but I'm wondering, as you guys are thinking

about this technology and all the different use cases across so many industries,

what do you think about the appropriate governance mechanisms that we have

at our disposal, the levers that we have?

The US isn't an island when it comes to this stuff.

A lot of other countries are racing ahead.

But I'm wondering, as you've kind of looked across the board here, if you feel like,

you know, there's any jurisdictions or even any companies

that are trying to get the balance right.

[Brian Williams:] Yeah.

So Anthropic is known as being relatively focused on AI safety.

And their tool, Claude, is actually one of my favorites to use.

I used Claude a lot when helping with actually write the scripts for Gen AI 101.

I write those with AI, which was great.

They're definitely a leader there.

And I think that, to your point, it's hard for our standard levers of policy

to keep up with the pace of change.

And so, in my opinion, to the extent that the users of the technology can just be informed

about its strengths, about its weaknesses, about -- I've been in many conversations,

I'm sure you both have, too, where people actually don't realize ChatGPT can be wrong.

They just assume it's always right.

And so letting them know about that, it can make mistakes.

Sometimes those mistakes can be very harmful.

I think that's a great starting point.

[Anne Ottenbreit-Leftwich:] And I think one of the things that I know we both do a lot, Brian,

right, is we really interrogate our AI, for lack of a better term, to say,

"Tell me the truth," like, "Give it to me."

"I want to see sources."

"I want to see what you're basing this on."

And then also even asking it to critique our -- what answers and what we've come up with,

or even our prompts of whatever that may look like.

So -- and actively asking for alternative ideas, or my favorite is to ask,

"What are the unintended consequences of this kind of decision?

And can you look at a wide range or myriad of different ways that we could approach this?"

So just think about that as if we asked for all of the policies that we put forth

in our local government, and etc., and said, "Here's our context.

Here's what we're looking at.

What are the unintended consequences if we put a policy like this in place?"

Right. Because there's always something that is impacted

that you haven't really thought through.

So I do think that part is critical.

I also do think that there needs to be --

and I know that there are some companies that are looking at really trying to make sure

that they're building things ethically or whatever that might look like.

I do think that universities have a really great opportunity here

to help influence large language models and try and think about what data is being informed.

And although at IU, we're using IU Enterprise tools in order to make sure that our data stays

at IU, I also sometimes will use public tools because I do want the model to be trained

on what I am giving it because I do think that a lot of the things that --

the way that I interact or whatever it is that I'm doing

or having my curriculum uploaded there, the different NSF curriculum that we've done,

I want that to become part of the model because I want it

to influence what other people are going to get.

[Scott Shackelford:] Mm-hmm.

Absolutely.

Absolutely.

This got me thinking -- and, you know, I know this is one of the prompts for the creation

of this Gen AI 101 course as well -- about what it means, right, to be AI literate.

This is October.

It's Cybersecurity Awareness Month.

As we're recording this, we've done a lot of work

around cyber hygiene at IU, digital citizenship.

AI literacy is an increasingly important component of that, right?

So from your vantage points, what skill set does that really look like, right?

[Brian Williams:] Yeah.

I think it can mean a lot of different things.

How I think about it is I think about knowing how to use tools and knowing their strengths,

which are maybe stronger than some people realize, but also the limitations,

which are also, I think, stronger than some people realize.

So there's a spectrum of what people think about AI.

I personally do not think we're anywhere close

to having a general artificial intelligence that's a true human replacement.

I think we're a long way away from that,

but I do think these tools can really assist a human who uses them well.

And to Anne's point earlier, one of my favorite ways to use them is

to make my own thinking better, right?

To try to come up with ideas I wouldn't have come up with in the first place,

find holes in my ideas, improve my ideas.

And I love that because one thing we emphasize in the course is the human as --

being the human in the loop, right?

So you're in all the system, and you can, you know,

critically evaluate what it gives back to you.

And so that's a really, really nice way to use it.

At the same time, also understanding, like, it does make mistakes.

It can get things wrong.

Lots of times, like that it -- how well it does, does it live up to the promise in the demos

that are shown, and by tech companies, right?

So you really have to be very careful about that.

I could definitely see kind of a world or think about the future of work,

where people have AI assistants, right?

You're always working with an AI assistant, but you're in charge.

And maybe you're managing a team of different AI assistants

that are specialized to different tasks.

But there's still, I think, a really, really important role for that human.

[Anne Ottenbreit-Leftwich:] Yeah.

And I think it really boils down in when we're talking about university students.

And I love the point, the human in the loop.

And I'll just say, like, Brian is also my human in the loop, so as I come up with things

or as we do different elements, like I'll often reach out to him or another colleague,

Justin Hodgson, to say, like, "What is your read on this?

What do you think?"

And so one of my colleagues often says, like, "This was developed by me and an AI."

Right? Like, so it's different iterations of whatever that might look like.

But I think for our university students, it really comes down to, like,

three main areas around, first, that critical understanding

of how AI works and what it means, right?

So that students can explain how AI systems generate results, identify potential bias

or misinformation, and recognize when AI use is appropriate.

The second is being the responsible and transparent use, which we talked about.

So this notion about students being able to responsibly integrate AI into their learning

and professional work, while also maintaining academic integrity and accountability,

because I think that's going to be a huge piece when they go out there.

And then this creative and collaborative applications, because it's not just

about critiquing; it's really about, like, what Brian was saying, this co-creation piece,

where students can combine human insight with AI capabilities to design new solutions

and generate ideas and advance domain-specific goals.

So I really think there's all these different pieces, and Brian has broken it down so lovely

in the Gen AI 101 course, really talking about using the prompt engineering pieces,

and then thinking about it as a thought partner,

and then thinking about it as a productivity amplifier.

So that's how he breaks it down in there.

And I think you have 20 different -- is it 20 or -- yeah.

[Brian Williams:] Twenty.

Yeah. Twenty skills that you'll learn in the course.

[Scott Shackelford:] And that's, you know, I think that's a useful segue to the next question

as we kind of conclude here in the next few minutes.

For, you know, listeners who might feel a bit overwhelmed or skeptical about AI,

I'm wondering, you know, what's one step or a simple first step at least

that you can take, you know, to engaging with it?

And, you know, assuming that it might involve Gen AI 101, what do you hope, right,

that people walk away from after completing that course?

[Brian Williams:] Yeah.

So I would say if anyone feels overwhelmed by AI, I'm with you.

I also feel overwhelmed.

I feel like it's impossible to keep up with all the advancements

that are happening all the time.

That's part of why we created Gen AI 101.

So it gives you a starting place.

The course assumes no knowledge of AI.

You don't need to have ever even logged into ChatGPT.

And it takes you from no knowledge to building your own custom AI assistant

in about three hours.

So I'd highly recommend you take that as a good starting point.

And one thing I hope people get out from the course is just that I think, at its best, AI --

it's not a replacement for humans, but it can be a really helpful assistant.

So both Anne and I have just heard stories over and over again from so many people

about how they've been able to use AI to help in their own life, whether it be at work or school

or home, after taking the course.

So that's been super encouraging.

[Anne Ottenbreit-Leftwich:] Yeah.

And I think the easiest tip is when in doubt, ask the AI, right?

Ask the AI, "I want to make a really great prompt.

Can you help me write a really great prompt?

Here's all of my information."

And the more information I think sometimes that you include, the better, because it'll suss out

and figure out which things are important or critical.

Or even asking it, like, "What are some different ways that I might be able to use,

you know, your particular system?

Give me some ideas.

Here's my job.

Here's my situation.

Here's what's happening in my life.

Give me some -- " you know, again, not oversharing, especially if you're

in a public one, because that's going up in training the model.

Although, again, maybe you want to train the model.

But again, I think asking the AI, "What can I do?"

"Help me rewrite these things."

And then even asking for specific videos that you might be able to watch.

It's a great recommendation-type of tool that you can use.

[Scott Shackelford:] And maybe just one last fun question on the way out.

If we are having this conversation five years from now, 2030, or even 10 years in 2035,

how different do you think this will be, the world will be, frankly,

in our segment, higher ed is going to be?

Get out your crystal ball and we can set our calendar reminders for 2035.

But I'm curious, you know, there's a lot of pessimism around the world.

But what do you think are some of the best-case scenarios?

Or what are you looking at in terms of the transformational impacts that are possible?

[Anne Ottenbreit-Leftwich:] So I think that the conversation is really going to shift

from this using AI to more designing with intelligence.

And so today, we talked about, like, prompting and policy, five to 10 years.

We're not even going to worry about that.

It's kind of like the one-to-one devices that we -- when we started worrying about that coming

to campus or Wikipedia, even further beyond that, right?

So faculty and students are not just going to use AI.

They're going to design, like, adaptive systems and curate AI behavior, thinking more about,

like, cognitive design, where we're intentionally structuring dialogue.

I think the bigger piece is more about identifying the importance of humans

in the loop, and where human interactions are critical, and what humans are best at doing.

So for me, with higher education in particular, or even K-12 education,

the most important thing is the relationships that you form with your students.

And a lot of times, there has been so much additional administrative burden that gets

in the way of doing those things.

We know time and time again, the thing that impacts student success is the relationship

that they have with their teachers or with their faculties.

So, trying to figure out ways to capitalize on that, whether it's experiential learning,

whether that is identifying different ways to offload some of those more laborious tasks

to the AI systems, and then figuring out what's next.

But I can't wait to hear Brian's response to this one, though.

[Brian Williams:] Yeah.

I think it'll be integrated in a lot of things.

And my hope is that it gets viewed eventually kind of like the calculator is

or how Microsoft Excel is, right?

I'm in accounting, Microsoft Excel came in, it was a big change, right?

But at the end of the day, it made accountants' lives easier, and especially the accountants

who knew how to use it well, right?

And so my hope is that Gen AI could be something like that, where it takes humans away from some

of the monotonous stuff and lets us do more fun, more exciting things.

And I'm with Anne, too.

I think it's going to be integrated everywhere.

And I think we'll see a big shift to voice AI, right?

We don't integrate -- we don't use MS-DOS anymore, right,

with text inputs everywhere, right?

Humans like to speak and like to interact in different ways.

So I see it going that way in the future.

[Scott Shackelford:] Excellent.

Sounds a lot more upbeat than SkyNet.

Let's go for that.

Thank you both, Dr. Brian Williams, Dr. Anne Leftwich.

Thank you for helping us unpack some of the most important technologies here

of our time, namely AI.

If you're watching or listening and are interested in registering

for IU's Gen AI 101 course, head to myiu.org.

Until next time, We Are IU, and we'll meet again.

Thank you.

Thanks for joining us for another episode of We Are IU.

Your home for fascinating conversations with thought leaders, legends,

and visionaries from the university we know and love.

New episodes available monthly at myiu.org.