(Music) Hi, my name is Nick Gasper.

I'm the director of online and digital education here at the University of Michigan, Flint, and I'm happy to talk about what artificial intelligence is and specifically what is generative AI.

artificial intelligence is a much larger concept where, you essentially are asking a machine or a robot to perform some sort of observation, analyze something, and then perhaps complete a task for you.

Now, how does that look in the real world?

Generally a regular just I would be say you're scrolling Facebook and an artificial intelligence is watching what you like.

It's watching where you're scrolling.

It's even watching what you pause at to see what are you engaging with.

And then what happens then is that it, decides what's the next post you're going to see?

Similar for YouTube.

It's watching what you watch.

And then making recommendations.

Spotify TikTok is doing pretty much how this works.

It's working within a boundary of a specific task or generative AI comes in.

It's just a subset of that.

But the difference is, it's more creative.

It has learned based on all of our activity and knowledge we put on the internet.

But more specifically is that it's going to generate something brand new based on what it's already learned from you.

when I talk about AI and generative AI learning, what I mean is that, we set up these systems to look at giant quantities of data, and typically it's the data that we put on the internet, and then we program them to analyze that data and try to understand what we humans are either communicating or what knowledge we have given each other.

And then it, and then it generates something based on what we've done in the past.

Now, what that really means is, you take the AI, it looks at a bunch of data, and then what happens is that these companies will pose a series of questions and then give it a series of answers and then have the artificial intelligence sort of, you know, quote unquote, reason its way to the answer.

So that is how it actually learns in the real world, is that it's sort of it's not sort of guessing.

It's actually trying to understand how we get from point A to point B within the data.

how would you actually use generative AI like ChatGPT to perform a task?

Imagine you're going to host a party soon.

And you're gonna send out invitations.

We have Easter coming up.

So what you might do is turn to ChatGPT and say, help me write an inviting note for my invitation, and then I'm going to send my family.

And then ChatGPT will write that note for you.

But what's interesting about this is that you can then go back and ask for edits.

You can ask for different versions.

So really the power still lies with you as the human, to tell it what to do and what to generate.

So, so that is one everyday use So if you wanted to use one of these tools today, the most common way you would use it is you would go to your favorite app store, and whether it's on your iPhone or your Android device, and you would download ChatGPT you would download Claude, or you would download Google Gemini.

Those are some of the biggest players right now in the AI space.

And then what would happen is that you would have that app on your phone and you would type back and forth with it.

That's the most common way you would use it.

But there are other ways, Google is integrated into all its phones.

Apple's integrating it into all their phones.

And that would be slightly different because now those tools you wouldn't necessarily need to download an app anymore.

It would sort of be integrated completely while you use your phone.

So you wouldn't necessarily know that you're always using an AI tool when you're working with your phone let's say you've downloaded ChatGPT and you want to start asking it questions and have it create something for you.

The best way to think about this is that hidden in that AI somewhere is an expert, and it's our job as the human to sort of find that expert so that they can help you complete whatever task you want.

And you do that by telling it three things.

Number one, who are you as the author.

And I don't mean personal information.

I just mean.

Hi.

I'm the director of an online department, and I need help, summarizing a report.

That would be me as the author.

And then you would tell it your audience.

So.

And if I had an example where I'm writing an invitation, I might say I'm writing invitations for my family and close friends.

So I'm telling it, who will actually be reading the content?

And then finally, what is your purpose?

So it's, it's the writing of the invitation.

It is maybe some complex concept, that I need to learn about, by asking you about that, maybe, help me understand how I charge, an electric car.

Right?

You're giving it a very specific purpose.

The best way to get the highest quality content that you are expecting, is by being very specific about what you're asking it.

Because you have to remember, this is one of those technologies where, it's what we call non-deterministic, where you don't know what you're going to get every time that you ask it a question.

But the more specific you are, the better chance you're going to get what you expect So let's say you ask it the same question twice.

You won't necessarily get the same exact response twice.

So in my example of charging an electric car, what will happen is that it'll it'll likely give you the correct steps each time.

But how it actually explains it, like the actual language it uses and the words it chooses will be different each time.

Generally speaking, though, the underlying message will be the same, but how it gets you the information will change.

at its core what it's really good at are a number of things which consist, generating text for people.

Just asking you a question, asking you to do something and just write something for you.

It can generate images, so especially, more advanced tools, more advanced versions of image generation tools can create very lifelike images that can fool almost anybody about whether they're real or not.

It's been advancing quickly, but it can also generate video.

Now, Sora is one of those tools to generate video, and it can even you tell it if you want a song.

There are tools that will generate any type of music that you would ask to let's talk about why this is really important for everybody to understand.

I was coming of age in the 80s and 90s when the internet came around and the internet advanced rapidly, very rapidly throughout the late 90s.

And we I now find or we now find ourselves in a situation where this technology is advancing maybe faster than the internet did in the 90s.

In fact, every day when I, when I read the news and try to stay up on on what's new, I find as someone who works in tech, I find trouble just trying just trying to stay updated on all the advancements that are happening weekly.

Monthly.

So that's that's why I think it's a number one that's important to understand what this technology is and how it works.

Because it's already being used in ways that you may not even realize yet in your everyday life.

So when you compose an email on your phone and you get a little like suggestion on how to finish the sentence, right, that's an AI helping you do that.

Or, let's say you're at the doctor.

Now, they should disclose this, but if they're typing notes on a computer, they could have a tool that's going to help them summarize that information, and then reference it later at a future appointment.

Or in the state in the medical world, we're now seeing advancements in artificial intelligence helping doctors identify medical images.

These tools get trained on millions and millions of, say, MRI eyes or X-rays, Cat scans and why we still need the humans involved, obviously, to make decisions about medical, health care.

The AIS are now helping doctors identify issues that can come up.

So, an example might be if you're having a cancer screening and there's an image, an artificial intelligence, if it has advanced training, might spot just a tiny pixel change between two images and say, hey, you should watch this to the doctor.

Like, give it a recommendation.

And on top of that, we now have tools that, you know, let's say a doctor is looking at that image.

And I could scan and query medical journals at a pace no human could ever match and go, oh, there's this obscure, you know, there's obscure case in this journal over here.

You just know about it.

So you can help the diagnosis for the patient.

So that's one of the.

So it's already happening in the real world.

But the other thing to think about and why this is so important is just how it's going to impact society We're now going to enter an era where knowledge is accessible to so many people now, where you can ask questions and get answers.

And, yeah, the technology, it does what we call hallucinate, where it just makes up an answer.

But that is happening fewer, as the technology or less frequently as the technology advances.

That's one of those things that these tech companies are really trying to solve, so that we get more accurate information from these tools.

But the other thing to think about, as it impacts society would be, say, the job market.

let me help connect this for everybody.

We are sitting in Flint, Michigan right now, and we could walk two streets over and you can go visit the Durant Dark Carriage factory.

And right.

We're building carriages.

But technology was changing.

Those guys, they they meet, right?

They meet a man named Buick, and then they go on to make General Motors.

And the thing to remember is that the people that were working in that factory were probably thinking about this horseless carriage.

It's coming around.

I'm going to have a job tomorrow, right.

But then what add up happening is that throughout the 40s, 50s and 60s, Flint, Michigan per capita was one of the wealthiest cities on planet Earth.

And so I'm going to say that I think about, one of my one of my colleagues said in a meeting.

So, Carson, wait, he said it's it's easy to think about the jobs that are going to be lost.

It's nearly impossible to imagine the jobs that will be created.

at this stage it's not an AI that's going to take your job.

It's going to be somebody who knows how to use the tool that will take your job, and that person will take your job in to others.

And that's why I like to advocate for everyone to understand how this works and how to get the best out of it and whatever task that you're performing.

Because if you're trying to enter the job market, or maybe you're trying to upskill, a lot of these corporations that are making big investments in the technology, you're going to want to make sure that you know how to use it and will use it to accelerate your own job.

so when you're using this technology or you're getting interested in it and you want to sort of venture in, there are some things to be really aware of.

The first is what, we like to call fleece wear in our office.

So that would be where you go to that app store to download, say, ChatGPT.

You could you could really just go to go in there and just type in ChatGPT you could do it for yourself.

And then you're going to notice that not only does ChatGPT get listed as an app, but you're probably going to see 50 or more other AI apps that will be intentionally made to look like ChatGPT, but aren't.

And these apps what how they work is that, they'll create a fancy little interface for you, and then when you type into it, it'll leverage chat, like it'll go out and actually ask ChatGPT and then return the answer to you.

But the reason you want to be wary of this is that, these apps can't.

They're there.

They they're coming after the your data on your phone.

They could come after the data that you type in.

They could even, say leverage.

Different alerts and notifications to try and get you to pay money that you shouldn't have to pay.

So that is probably the biggest, the biggest suggestion I could make is when you download or use these tools, make sure you are using a reputable tool from a company like OpenAI who makes ChatGPT, or Microsoft, who makes, Copilot or Google that makes Gemini.

The other thing to know is that you start using this, and maybe you start looking around on social media that this technology has the ability to create a lot of misinformation very quickly, right?

If it's used incorrectly.

So when you are scrolling through, say, your social media feed, be very careful about when you look at images.

Sometimes those images aren't even real at all.

There are some tools in the in the early days, say in 2022, that we would say look for hands with more than five fingers or, maybe the shadows on an image don't quite diverge the way you would expect, but nowadays it's far more difficult to know what's real and what's not.

So my suggestion would be to look at, this is the skin on somebody a little bit too perfect or, you know, maybe the buttons on their shirt don't quite line up the way you would expect.

And it's much, much more subtle now.

But the other thing to be aware of is that the things that you read as well, they could be written by an AI tool in less than five seconds.

So always try to read things with a discerning eye.

You know, I think about my own childhood and we we used to get we were told, you know, somebody from the millennial generation, we were told, don't believe everything you see on TV.

I think all of us need to remember to not believe everything you see on the internet, because it could be anybody writing it, including about these days.

So if you're interested in finding more information, I would recommend, some resources here at the university that we can provide.

They're freely available to the community because I'm a big fan of spreading that knowledge around so that everybody, can protect themselves or and also learn about how to get the best out of them.

So, I would recommend you go to dot.

Flint.edu/jen, I, and that we create a whole host of resources that anybody inside and outside of the university can, can leverage and read and, I also have a course that I've created.

So when we were first exploring this technology, I noticed a growing gap between, people that understood how to talk to these tools and get the best out of them, and people that didn't and the people that weren't getting the best weren't using them and learning about them.

And so I created a course on artificial intelligence literacy that is available to everybody inside the university, outside the university, the general community all around the world.

So we currently have a little over 2000 people enrolled in it.

And there you will learn, you'll learn about how to get the best out of that.

You'll learn what it is at a more detailed level, and then you'll learn how do you actually talk to these tools to get high quality content from them?

And I, you know, I really hope that, everybody, even in the community can join us in that space to learn more about it.

So to wrap things up, I hope that everybody will remember that this technology is really just it's really a tool, but it's a tool that can really have benefits to society in our community.

It can hopefully accelerate education.

And through the growth of knowledge, it can prepare us for the coming job market.

And I think about, you know, what is the best way to help somebody, pull themselves out of poverty?

It's education through and through.

And then if we can get this right, we can really help accelerate that for people and really make our communities a better place.