- [Voiceover] Live from your public media studios, WVIA presents Keystone Edition Business, a public affairs program that goes beyond the headlines to address issues in northeastern and central Pennsylvania.

This is Keystone Edition Business.

And now moderator, Steve Stumbris.

- Hi, I'm Steve Stumbris.

Artificial intelligence or AI is making waves across the professional and educational landscape.

Some people are afraid it will replace them at work.

In some fields, it's already transforming jobs.

Many are studying the cost-saving and time-saving benefits.

How will AI affect you and your future?

If you have questions, you can email us at keystone@wvia.org or tag us on social with the hashtag, keystone business.

But first, Sarah Scinto goes deeper into what AI is.

(logo swooshing) - [Sarah] Some call the advent of AI or machine learning the next industrial revolution, but some worry about what it could mean for society.

What exactly is AI?

AI stands for artificial intelligence.

It's essentially a machine's ability to do something that would typically use human reasoning.

Ai text generators can answer simple questions or create longer written works, like school essays or speeches.

Image generating AI can simulate certain art styles or make empty theaters look fuller.

But there are some ethical questions that come with these tools.

Is using a text generator plagiarism?

Some AI image generators have been accused of stealing art from human artists who put their work online.

One danger of AI is misinformation.

It can create convincing images and videos that can look and sound official.

Another issue is the lack of differing perspectives.

Only a third of people developing AI right now are women or people of color.

The concern is that a lack of diverse input leads to a homogenized output.

And what AI generates is seen primarily through one kind of viewpoint.

On the other side, AI can speed up tasks, like data collection, and it can do repetitive jobs without getting bored or tired.

There's a lot to learn about AI as it continues to develop.

And there's more information on wvia.org.

For Keystone Edition Business, I'm Sarah Scinto.

(bright instrumental music) - ChatGPT is one of the most common generative AI tools.

It delivers text based on a question or prompt.

Here's how quickly that works.

Give me three quick bullet points summarizing what business owners should know about AI.

And at about the time that I've finished speaking my prompt, ChatGPT has given us our business owners' three things to consider.

I'd like to introduce our panelists who can share their experiences with artificial intelligence.

Thiago Serra is a leading researcher in artificial intelligence from Bucknell University.

Alex Fleck is a serial entrepreneur and marketing consultant, as well as founder at Zenith Online Marketing.

And Lindsay Fairman is the Chief Technology Officer from TEAMology.

a company using AI in the mental health and education space.

Well, thank you for joining us.

I'm looking forward to the conversation today.

And for viewers, if you have any questions, please ask at keystone@wvia.org.

- Thiago, welcome.

Alex, welcome, Lindsay.

So pleased to have you joining our panel tonight.

Looking forward to this conversation, really interesting topic ahead.

We heard in Sarah's opening about what artificial intelligence is, but Thiago, I wanted to ask you first to expand on a couple more terms that are very common, a language model and the idea of generative AI.

Can you tell us what those mean and our conversation will build on that.

- Okay.

- So, well, as was mentioned before, artificial intelligence has been around for half a century now since World War II, but what has been different in recent years is that we're using more and more of machine learning, which is this idea of learning from a lot of data.

And more recently, we started working on using machine learning and using all this data that we have been collecting in many different ways to generate outputs, right?

So you have all this data, for example, songs, pictures, plays, books.

And if you understand the patterns in this data, maybe you can replicate them.

And what we see from large language models now is a machine, a mathematical model that by analyzing all this data that mankind has produced, humans in general, not only men, is able to produce something similar.

And suddenly that became very, very productive.

It's more... What's the right word for that?

When you look at it, it looks very good.

It didn't used to look that good in the past.

- Tools have improved so rapidly and so rapidly within the past year alone, it seems.

Is there some sort of step change?

What made that difference happen?

So AI's been around for decades.

You've been researching and teaching and studying it.

What happened in the past year that really made it take a step forward?

- Well, it has been a long process.

For sure there were many points where we call, there were AI winters where research in this area suddenly stopped, or there was no funding for research because we got to some roadblocks.

But I believe from 2015 to now, suddenly the spike of development was strong.

And I feel that there was this one paper called "Attention Is All You Need" in 2017, where they finally created a mechanism which is what we call attention for understanding how, in a context of many different pieces being given to like a machine learning model, we can understand which ones are more relevant.

For example, when you write a prompt, what are the words- - Yeah.

- in your prompt and how to connect them to understand what really the user wants?

So these attention mechanisms have been around since 2017.

And so in this five years until end of 2022, this led to a sequence of developments that suddenly brought us to ChatGPT.

- This sounds very similar to many other fields of endeavor where it takes a lifetime of work to become an overnight success.

So Alex, I wanna turn to you now about how you are using AI- - Yeah.

- in your day-to-day life.

We, as the general public view AI systems as incredibly complex.

It's built on years and years of research and practice, but you are using tools like ChatGPT to generate new business ideas, as you develop new business models and launch them.

Could you tell us a little bit about a day in the life of an entrepreneur and how you are using AI tools nowadays?

- Yes, absolutely, Steve.

Thanks for having me as well.

Yeah, so I think in general, what we face as entrepreneurs and business owners is we're always trying to do more with less time.

So we wanna launch new businesses, new product lines, new creative enterprises, whatever we're trying to do.

But the problem we face is there's only so many hours in a day.

And so we always come up against this thing of, well, I have to pick what I can do.

And there's maybe things I'm leaving on the table that I would love to do, but I just don't have time.

And so what we found is, and particularly in the business field, and I think this really took off quickly over the last 18 months, year or so in particular, is that we can use it for all sorts of facets within the business, from ideation to market analysis, to competitor analysis, to pricing structures, profitability metrics.

We can think about content, so curation of web content, social content, how we go about creating a brand voice, a logo.

Ideas about font matches we've done when we're building websites.

So we've really, and I in particular have really installed it in every facet of the business from that early ideation right through to everyday operation.

- I understand from you that your most recent business that you launched, how ChatGPT was, and here's a term that Thiago used before we got started, your "co-pilot and assistant."

- Yeah, yeah.

And that's the perfect analogy for it.

I always call it my digital ally, but it's the same principle.

It's always there.

So whenever I'm stuck or I feel overwhelmed or just brain fog or it's not quite, you know, the inspiration isn't quite there, I'll just turn to part, you know, predominantly ChatGPT, but other AIs as well.

So when I find myself stuck, I need progress fast, that's where I go.

And it sort of tends to open the ball.

You know, you kind of get ideas flowing faster than ever.

When you are stuck, they come, you know.

You can then just use your critical analysis to determine, yes, this is good, let's go down this pathway, or let's try something different, try a different approach.

- Well, thanks.

So Lindsay, I wanna bring you into the conversation here.

We heard from Alex about how he, as a serial entrepreneur is using AI.

You are in the role of chief technology officer for a growing company in educational technology.

Wanna ask you to describe how AI is allowing TEAMology to grow as a company, how you are using it to develop your services, and what should, say, educators and parents who are hearing about TEAMology, what should they be excited about that AI is allowing you to bring to the market?

- Yeah, thank you, Steve.

Thanks for having me tonight.

And so at TEAMology, we are focused on strengthening the social, emotional and mental health of children in grades K to 8.

So we're working with young children, really trying to help give them almost a sense of mental fitness to strengthen their ability to have coping skills and strategies throughout their life and just have a stronger foundation for their mental health and mental fitness.

And so in order to do this, we're really focused on overcoming two challenges right now happening across our education system.

First, we have an increase, unfortunately, in mental health needs across our children, even down at the youngest ages.

It's often assumed it's occurring at higher age groups, but we are seeing it in elementary, and it's a challenge that we need to really start addressing and not waiting for the children to grow older.

And then we're also focused on, there's a national shortage of counselors, teachers occurring in the education system.

And especially for counselors, right now there's an average of 1 to 500 students on a counselor's caseload.

A counselor cannot see 500 students in a week, in a month, in multiple months.

And so that's the challenge that TEAMology is focused on overcoming.

And with that, we're using technology, we're using artificial intelligence, and we're using humans to overcome that.

So the way that we use artificial intelligence, because one of the things that we are very focused on are our ethical principles for AI.

And I saw that as one of the three responses in your chat prompt, right?

That's really important to understand the ethics of the company that you're working with.

And for us, we are not outsourcing children's mental health support to artificial intelligence.

We're not at that point in society, we're not at that point as a company and we're not relying on artificial intelligence for that purpose.

What we're relying on artificial intelligence for is to, (coughs) excuse me, look at the conversations that are occurring with our students and to look for patterns, just as Thiago described, looking at historic patterns that are raising signs of concerns and we're using artificial intelligence to bring those to the surface quicker and easier with more clarity for our team and for the school team to recognize that there's signs of maybe some bigger concerns here that normally would not be perceived by an adult through a conversation.

This is considered a reactive model.

So with looking at historical trends and patterns and applying those to what the children are saying throughout our system, and helping to bring this to light quicker, which is helping us to then raise concerns to the school quicker, (coughs) excuse me, easier, and in a method that's meaningful to the school.

So it's not deep in technology.

It's not deep in languages that aren't commonly used in the education system.

We're able to apply it in a meaningful manner that then the school can prioritize who needs attention quickest.

And so through that, artificial intelligence is helping us to be faster, more efficient in raising this to schools.

- That seems to be a common response that I've heard from practitioners, from people who are using AI, it allows them to be faster and more efficient.

I wanna turn back to Thiago and ask, you're hearing about Lindsay and her company are using AI to address a shortage of counselors and giving them tools that help them be more efficient.

Can you expand to other industries, other types of jobs where AI already has taken root?

Or perhaps in the future, what types of jobs are going to change the most because of accessible AI tools?

- Well, we're certainly seeing AI being used for content generation, right, which as Alex mentioned, helps him a lot.

And so in a certain way, if you're going to be working on content generation, you're going to be overseeing a lot of content being generated and you're trying to be doing the oversight of that.

Like what are these co-pilots or assistants are doing for you?

In software engineering, we can see the same thing.

We do have some tools that have been available for much longer than ChatGPT, for example, GitHub Copilot- - Okay.

- where the idea is you need to write code to, like, you need to write software that's going to perform a certain task.

And as you're writing the lines of code, there's a repetitive pattern there.

Because a lot of software looks very similar, the next lines are going to be suggested to you in the same way that today when you open a word processing like software, like Microsoft Word or Google Docs, words or sentences are suggested to you.

Sometimes they're going to be right, sometimes they're going to be wrong.

It's too important that the students have the fundamentals to be able to discern what works and what doesn't.

So we do need to have professionals that have the depth to understand when to intervene because of how these tools work, but certainly a lot of content, in many different ways, is going to be automatically generated.

So we can't control that in so much detail anymore.

To a certain extent, I feel that this is very similar to talking about calculators and engineers.

Well, if you have calculators, you don't need the engineers to do the math, but they should know the math, otherwise it's going to be dangerous.

Especially if I don't, let's say, you start writing software that does more complex calculations, you should have someone having a feeling, does this look right or not?

And with these tools, it's going to be very similar.

- Is this something you already see impacting the classroom experience, the experience of students in higher education?

What do you see?

- Absolutely.

Well, if students don't know anything about, let's say SQL, which is a language used for database querying, they can quickly figure out how to do certain things using ChatGPT.

It doesn't mean that they're going to do it perfectly, but they can very quickly get used to a new technology or new programming language or just get something done, having a certain background already.

I had students that reached out to me saying, well, I'm going to have an interview and I need to understand a little bit more how this language works.

And by the time I was available for the interview to talk with them before the interview, they said, "No, I just talked with ChatGPT and I figured out everything I needed."

- [Steve] Yeah.

- So they're becoming more productive and students have been faster than, let's say faculty, to see the potential of this tool.

For one thing, when we think about how we're going to prepare them in the future, we do have to stress to them that they should still know the fundamentals.

They cannot rely on this to do their work, otherwise they're going to be replaceable.

But for sure they need to know how to leverage this in the same way I cannot ask them for things that are easy and predictable to do because then they can just ask ChatGPT to do these things for them.

So it's fast pace.

It's a fast-pace change.

We're all seeing this moving around below our feet right now, but education theory is important because you need to know how to do the creative tasks, the ones that we cannot find from the patterns and how to coordinate all of those things, right?

- Yep.

- Lindsay, as someone who's also involved in education as TEAMology provides services to teachers- - Yeah.

- and in school settings- what do you perceive about how this is changing how students learn, or how students, as they become the employees and the business leaders of the future, how's this changing things for them?

- Yeah, just to really, to piggyback off what Thiago just shared, I just read a really interesting article in "Harvard Business Review" about a concept that's being called the "Citizen Developer."

And this is the rise of, I'll say kind of your general knowledge employee that's using ChatGPT to generate code.

And they're deploying software in environments where nobody around them really understands what that software is doing.

And it's talking about the problematic raise of software being put out into the world that the team really doesn't understand.

I think this is a great example of highlighting how ChatGPT is helping employees, professionals be more efficient, but the gap is going to be in understanding what's under that foundation, what's behind the work, what the code is really doing.

And in this example, when the software breaks, people can't fix it.

They're having to rely on ChatGPT, on artificial intelligence to fix it.

And so where I would say the biggest kind of employment opportunity in the future would be is the person that understands how artificial intelligence is made, the person that understands.

And I'm a tech person at heart, computer science, engineering, all those great things.

understanding what's behind it, what's behind software development, so that you can be creating artificial intelligence instead of having it feed your job.

I thought that was a fascinating example of the impact of kind of the future of the professional.

- Lindsay, more about your background in computer science and as a developer, what do you hope for the future of AI tools with a more inclusive range of society being involved in understanding and developing those tools.

We heard at the opening of possible bias or explicit bias that is presented by some tools because of how they are programmed or what data is put into them.

Can you expand on that?

- Yeah, I just kind of, from my personal perspective, I would say now we're seeing more of the impact of the collision of culture and society with technology.

Before really kind of this introduction, a lot of of technology was, almost your basic ones and zeros, right?

You have kind of your fundamental algorithms or software or technology that's designed to complete a task.

But now we're seeing more of the integration and consideration of bias, of culture, of society, where you're not just looking now at the result of the product, the result of the engineering process, you're looking now at who's behind it.

- [Steve] Yeah.

- And as I mentioned earlier, and I think it's a continued trend of who's the company, what are their ethics, how are they communicating this?

How are they documenting, back to what Thiago described, their model, their data used to create this?

All of that's gonna matter a lot more going forward than it has in the past.

- Oh, thank you very much for that perspective.

Alex, I want to turn to you, and I hope a takeaway from this overview of AI and the future of work is for business owners to start learning about what they need to know about AI and how it's going to impact their employees, their industries.

As you go about growing businesses and consulting with businesses, what are some key things that you're telling business leaders that they should know about AI and impact on businesses in Pennsylvania in the future?

- Yeah, for sure.

I think there's a couple of key points.

I think to build on what Thiago and Lindsay said, you know, it's critical that the operator, as in you or I, when we're using AI and ChatGPT in particular, know what we're trying to get out of it, we understand the reasoning behind it and we view it with a critical and creative mind.

So when we go to prompt ChatGPT to get results from it, we understand what we're really looking for, and we can look at it.

And I said to you before the segment, Steve, I was doing something last week, I was asking it to calculate some profitability.

So we're looking at figures of new pricing plans and so on.

And it just didn't look right when it came back to me.

So I delved deeper and I realized that it made an error early on in the process.

So things like that, at a base level, as a business owner, it can be very useful, but you have to have your wits about you, you have to keep an eye on it, and you have to learn how to leverage it in better ways, more efficient ways.

It's not just like a cheat code.

And it certainly is not gonna be a substitute, it's gonna be a supplement.

So it supplements our understanding of business and markets, you know, where as a solopreneur, you're trying to launch a business, you maybe don't know anything about the market, but within seconds you can have a load of data and real-time data as well because within ChatGPT, of course now we can use plugins to get real-time data.

It doesn't have to be from a couple of years ago.

And therefore, we can get all this real-time data, we can put together reports, diagrams, slideshows, slide decks, website structures, content, all from the course of one chat.

So I think as a business owner, it's really important to delve in because yes, this has been around for a number of years, yes, it's a buzz word and buzz topic right now, but I firmly believe that as a business owner, as an entrepreneur, if we don't dive in and learn about this now, in three to five years time as a business owner, we're gonna be left behind.

And so it's really important that we leverage the cutting-edge technology as it arises.

And don't be afraid to use it because it's literally, as you saw at the start, you go in, it's just like writing an email.

It's just like asking Google for something, except it's this iterative process of learning about your own business, about other topics, like Thiago had mentioned, you know, being able to learn faster, but still, you have to have that underlying understanding and critical thinking as you're the operator of this co-pilot.

- Good advice.

- Thiago, as someone who, hopefully, has a bit of insight into the history of AI and at you're at the forefront of contributing to and reviewing the most current research, the newest knowledge in this field, what excites you?

What concerns you?

What are you seeing out there as the next thing from your work that might have an impact on all of us as we encounter AI in the future?

- Well, if I'm gonna stick this in my direction.

what really excites me is how we could potentially integrate this artificial intelligence with decision-making with, for example, mathematical models that we can use to make better decisions for companies 'cause I can certainly write lots of equations that can describe how business work internally, but there might be unknowns that we can learn from the data that really excites me.

But switching to the concerning part, and I really want to tap on what Lindsay was mentioning earlier, we do have the ethical issues, and that has to do also with the fact that AI is not perfect.

It's going to make mistakes, but these mistakes are not going to be uniform.

And that's the problem that creates societal issues.

For example, there was this groundbreaking study some years ago by Timnit Gebru and Joy Buolamwini about how facial recognition systems would work better with all with like whites and men and worse with Blacks and women.

And the consequence of that is when used by law enforcement, you would have incorrect indictments of minoritized population.

So they would be more often mistaken by someone who committed the crime than someone who is men and white, which is part of the concern.

It has to do with the fact that we have more data on the majority, on, for example, white people and on male when we're creating these models because it has to do with who develops them and how developers understand, "Yeah, this is representative.

It looks like me," right?

- Yeah.

- which brings actually the beginning of the program.

And the other part that I should...

I think we're running out of time here.

So- - We're now.

So strong concern, amazing opportunities.

- Yes.

- And to bring us to the close of AI and the future of work, I'd like to thank you for joining us tonight.

Thank you for being a guest on Keystone Edition Business.

For more information on this topic, please visit wvia.org/keystonebusiness.

And remember, you can watch this episode on demand anytime online or the WVIA app.

For Keystone Edition, I'm Steve Stumbris.

Thanks for watching.

(relaxing instrumental music)