- This is a production of WEDU PBS, Tampa, St.

Petersburg, Sarasota.

[Music] - Coming up on Florida This Week, artificial intelligence is moving faster than ever, and Florida is having to figure out how to keep up.

From the data centers powering AI to technology being developed for the military.

AI is also changing how doctors detect disease and how criminals target their victims.

So where does AI go next and how do we make sure the technology works for us, not the other way around?

Our panel breaks it all down next on Florida This Week.

[Music] Welcome back to Florida This Week.

I'm your moderator, Lissette Campos.

Artificial intelligence is moving from something that once felt futuristic to a technology already changing communities, businesses, health care, and national security.

We've assembled an expert panel for you.

Joining us are Christian Casale, a journalist with the Suncoast Searchlight.

His reporting has examined the growing regional debate around AI data centers and the infrastructure needed to support them.

Dr.

Ankur Mali is an assistant professor at the USF's Bellini College of AI, Cybersecurity and Computing.

He is also the director of the Trustworthy, Knowledge Driven AI Lab.

Next we have Dr.

Irfan Ali.

He is the president and CEO of Pioneer Medical Group with more than 300 health professionals working in the region.

He's also the author of Fractured But Fearless.

And Sonny Parsons.

Sundanah Sonny Parsons, he's a supervisory special agent with the FBI's Tampa office and a subject expert on AI and cyber crimes.

So let's start with the infrastructure behind the AI revolution.

As the AI systems become more powerful, the demand for data centers, electricity, water, and other infrastructure is growing.

And that's intensifying a debate in Florida communities over whether and how to welcome AI data centers.

Christian, you've been covering this extensively.

What is the mood that you see out there when you're out there doing your stories?

- It's overwhelmingly negative, at least from the kind of average person.

You know, I was at a meeting in North Port, Florida several weeks ago.

North Port is one of the fastest growing communities in the country, let alone Florida.

I think it just crossed 100,000 people.

And the city commission there had a meeting to talk about an 18 month moratorium on all data centers.

This came after an application for a 50 megawatt data center to be built in the city, and there are about 400 people that showed up.

It was an eight hour meeting.

It ended at 2 a.m.. Of the 400 or so public comments, probably three were in favor.

Even those were a little wavering.

I think overall, people kind of have seen the worst of what data centers have to offer.

A lot of stories about how much they're taking up water, how they're jacking up utility rates, how there's noise pollution.

I think there's a general sense that AI might also be coming for their jobs.

That's not very good for society as they see it, and no one's really making the case for them of why it should be built in their backyard.

And I have met 1 or 2 people in all my time reporting on this that are in favor of a data center in their community.

- Cities and counties are not.

Some of them are just taking matters into their own hands.

And there are some moratoriums in place across the state of Florida.

Tell us about that... that movement.

- Yeah, sure.

So I think one issue is that a lot of municipalities, cities and counties didn't really have any specific development codes relating to data centers.

So when they got applications, they didn't really know what to do with them.

And Florida's development law is already pretty friendly to developers and those who want to build residential commercial developments.

So people that want to build AI data centers could take advantage of that because there weren't specific regulations on them.

So very quickly, you saw across the state, cities, counties putting 12 month moratoriums, 18 month moratoriums, trying to keep at least the largest data centers outside of their backyards.

But those might not hold a lot of water.

You know, I'm hearing from some county attorney, city attorneys that Florida law might not even let them kind of put this burdensome regulation on any kind of development, even AI data centers.

You know, a lot still kind of up in the air legally.

- What is the argument on the other side?

What do communities stand to gain from having these AI data centers?

- Well, I think an important point is I've covered residential developments before.

And typically when someone wants to build, say, a 6000 home development, you have a lawyer and a good suit, come to the commission meeting and try to talk through everyone's fears and let them know why this is good for the community.

Uh, you know, at least in the meeting I covered, North Port, the developer of the data center was not there.

He's out of state.

No one's really even sure who has the ultimate corporate power there.

They just filed the application and expected it to pass.

So there's not really a good PR effort, at least in this case, for someone to make the case of why you should be accepting this into your community.

You know, the overriding argument is that is property taxes.

You know, the property valuations on this land would go way up if a data center is built.

That's a lot more revenue for cities and counties.

And especially if Amendment 3 passes.

Cities and counties might be a lot more receptive to things that would boost their coffers.

- Amendment 3 would be the reduction in property taxes.

Of course, on the policy side, Governor DeSantis has taken a different approach than the White House.

Um, you know, what is the current policy that we have in the state of Florida?

He has offered up the AI Bill of Rights.

It didn't pass in the last legislative session, but these are things that he has offered up.

What are we currently doing in terms of policy.

- Yeah.

That's right.

So Governor DeSantis had kind of unveiled his AI Bill of Rights, I believe, six or so months ago at the beginning of end of 2025.

And, you know, that pushed for protection against things like deepfakes, more consumer protections for data centers to pay their own way on utilities and water and things like that.

That passed the Florida Senate did not pass the Florida House.

You know, the former House Speaker, Daniel Perez, basically said, look, there's an executive order from President Trump that says that states cannot make their own AI policy that has to come from the federal government.

And then Speaker Daniel Perez was appointed to the ambassadorship to Brazil.

- That's right.

So he's no longer the in the lead.

That's true.

- So we'll see if that holds.

But the federal government has been very clear that we're going to set the policy on AI.

States can't do it.

We don't want 50 different regulations on AI from 50 different states.

So right now, you know, the regulations for Florida are the same regulations across the country.

- So DeSantis is term limited.

So where do the candidates for governor stand, Byron and Jolly?

- Sure.

So Byron Donalds has been, uh, pretty pro data center throughout his campaign.

I believe he had one quote where he said, Florida has five extra gigawatts of energy on its grid, so we could build five one gigawatt data centers.

He's brought that back a little bit.

Since there's been a lot of criticism of data centers throughout this political season.

He's advocated for other companies to pay their own way.

As far as utilities.

Congress just passed the, I believe, the Ratepayer Protection Act yesterday in the House.

That would if you read that.

- That it's something that the company should look into but it doesn't require.

- Does not require it.

It says that they should consider a federal framework does not actually require them to pay their utility rates.

And that's something people are really worried about.

This election season is about affordability and Florida.

It's about overdevelopment.

And the data center issue kind of intersects both of those.

- And the infrastructure debate is happening at the local level in a very intense way.

But also, Dr.

Mali, AI is also becoming part of much of the bigger questions around national security.

Tell us about that.

The work that you all are doing at the Bellini College of AI, Cybersecurity and Computing.

- So as we know, it's being a foundational technology and it's almost used in a wide variety of sectors, if not all, but almost it's there, it's touching, uh, the door knocking and then saying that, okay, we are here.

So within USF, we have been working on cyber security hardware, uh, network security in health, um, even in terms of, uh, biology, additive manufacturing.

- In different business sectors.

- Different business sectors.

It's even used in law.

Like we have, uh, a faculty who's working, uh, with people in law to create these trustworthy aspects, right?

So one of the project, which we have been working on in health is more towards understanding that, okay, if we have, we have these AI technologies, it can understand the pathology report, it can tell you, okay, if there's a cancer is there, there's not, but is not trustworthy in nature.

So we are our team is building more like a neuro symbolic system, which takes information from clinician, their knowledge and integrates that with an AI.

So you have more clinician inspired AI where the symbolic knowledge is coming from clinician so they can trust the output of it.

It will give you a more explainable aspect of what this is.

- And we'll talk about health and AI in a moment.

But first I'd like to ask you about what you all are doing in terms of military defense.

In the spring, USF announced a very important partnership with a defense contractor, and you all are working on designing a GPS denied environments.

It's a project where you can, um, have that.

Tell us about that, how that will work.

- So this comes from a digital twin technology.

So digital twin is a simulated environment which approximates your physical world in a with a precision, right?

So you can test, analyze, see the test case scenarios for different AI agents that what they can, what they cannot do.

And this is a more interesting use case where imagine that there are swarm of robots, a swarm of autonomous agents who are doing some task, and they need to know the corresponding longitude and latitude information right where they are and how to coordinate their information.

- So if a navigation system is being jammed or that's where the GPS denied aspect comes in.

They have to be able to.

- They have to be able.

- To complete the mission.

- And in majority cases, what happens is they rely on some other data which can also be hacked in some way.

And then such unreliable prediction comes in, right?

So we are creating these tools where students are stress testing that, okay, if agents cannot communicate with each other, there's a hacking event, there's GPS denied or some aspect of robots.

Like if you have like thousands of agents or only 20 are working, then how to complete the mission in a more explainable or responsible way.

- And that's something that I'd like for you to explain, 'explainable' and 'trustworthy.'

That's something that is very much a part of your research and defense.

How important is that aspect of explainable AI.

- In the current era?

It's very, very important.

I have to emphasize that point because when we say trustworthy, it's not about developer trusting it.

The human who is using this technology should also trust.

What is the outcome of it?

He, he or she should be able to see that, okay, this is what AI is doing.

This is the rationale behind why AI is making that decision.

And they should trust the outcome of an AI agent, right?

That's where the trustworthiness comes in.

But the next thing which we are pushing forward is AI should also give you the scenarios where it is uncertain that it is not confident enough in its prediction or it has never seen that scenario, so it is not reliable.

And what is going to predict?

That's where the human comes in.

And then they can analyze and say that, okay, this is what it needs to do rather than blindly trusting what AI is.

- So we're going to go now from the battlefield to the hospital.

Dr.

Ali, I'd like to bring you into the conversation.

One of the areas where AI could have a direct impact on people's lives and in a very positive way, is in identifying disease and other risk factors at a much earlier stage.

Is there a real world example that you can give us?

- Thank you so much for having us.

The world is changing, as we all know.

I think that when you look at traditionally how medicine has been practiced, it was always been reactive.

A person is sick, they come to the hospital and we're trying to find out what is wrong with them.

But we the movement which is happening today is today's predictive medicine.

How we can figure it out?

Things catch them early.

Nobody is saying that AI is going to give you a diagnosis or going to replace the physicians.

However, it gives you signals early, and I think that is the biggest thing that we can get out of AI.

There are so many different examples, from pancreatic cancer to prostate cancer to lung cancer, that we are able to we have made strides in diagnosis.

But I think the biggest thing is that today, with all the imaging, with all the testing that we are able to not just see cancer early, but the biological footprint of a cancer, which is going to develop in 16 months from today, we are able to diagnose them or able to predict that it will be happening and we can catch them early.

- You recently presented these concepts to a health care leaders from around the world.

And what do you say to them when you point to the work flow changing?

What are the things that they need to get right before they deploy AI in their hospital systems?

- So this is a...this is a complicated answer.

The way I look at it, um, you know, it has these four these components.

The first part needs to be we need to have validation.

If we do not have the validation of the software, the AI tools that we are using.

One of the biggest thing is confidence bias.

Just because somebody is saying that it is great doesn't mean it is great.

- What do you mean by validation?

- Validation mean that the AI systems that we are using, AI is not sentient.

It does not have the consciousness that you and I have.

So it is designed to continue to regurgitate answers.

You know, it never says, I don't know.

Right, so the hospital systems and all of us need to think about the two extremes.

One, we love it because it is sexy and it is new and we will embrace it.

The other part is that it is new and unconventional.

We don't want to take it.

So the first part is governance.

The second part needs to be integrated.

I get 90 million alerts every day, right?

If it is not integrated with the physician's workflow, it will be of no use, right?

The third part of, of, of that is um, is the governance of this thing that, that what we are using the cyber security, your and my personal information, how people can get hack...can hack into these systems.

If we do not have those boundaries and if it goes, it goes crazy.

How do we bring it back?

I'm a big, big supporter of that.

And the last part is humanity.

I can figure it out.

I mean, medicine is not just about the CT result.

It is not just about how much MRIs I can get.

It is it is a relationship between a vulnerable person and a person they trust to treat them.

The humanity aspect of it will always need to stay.

So I think when we make the decision, these four things in my eyes are the forefront of, uh, of decision making.

- And Sonny I'd like to turn to you at this point and bring you into the conversation.

You know, hearing Dr.

Ali talk about the good things, the potential for saving lives and identifying disease earlier.

There's so much potential for good with AI in healthcare.

Uh, the bad guys are using AI as well.

Absolutely.

What is the FBI seeing in terms of the sophistication of AI in the cyber crimes you're investigating?

- So cyber crime is nothing new, right?

It's been going on for a long time and it's going to continue into the future.

Um, bad guys are smart as anyone else.

And they're going to use the new tools that are out for to meet their goals.

Great.

Uh, of most of which are financially motivated, right.

Um, so a big part of kind of one of the big things we're seeing is mass personalization at scale, meaning they're using these AI tools to scrape the web of all these databases that have been built on us and social media and everything, and then making these very tailored, personalized profiles of us to, you know, send out phishing schemes that are so believable.

Where before it was very rubber-stamped, you know, someone from Nigeria with spelling mistakes, I think we can all catch those now.

But now it's so complex and targeted and polished that it's hard to tell the difference.

So I think that's that's one of our biggest challenges.

- So the AI is very much a part of the reconnaissance and the targeting.

Absolutely.

The prep work, you would say, how are the criminal actors leveraging AI to enhance the traditional cyber crimes?

- Um, so while traditional cyber crimes, um, you know, are being enhanced, uh, I think with identity fabrication, they're using these tools to fabricate video, audio, um, pictures, you name it.

Yeah.

- The deepfakes you've talked about.

- All of that.

- The AI generated voices.

That just sounds like something out of a movie.

- Yeah.

Well, no, it's happening now.

And they're using that to, to trick people into doing things that they normally would not do.

Um, you know, whether it be sending money or, or gaining access into areas of organizations that they shouldn't be in.

So the human element of that is usually the failure, but AI is just a powerful tool enabling that.

- What challenges do are you all facing in law enforcement in terms of dealing with this AI generated evidence?

- So I would say the biggest challenge we face right now is authenticity.

What are we looking at?

Is it real?

Because that can determine someone's innocence or guilt.

And that's not a small thing, right?

Putting people in jail over.

We better be right on the evidence we're dealing with.

So, um, a lot of these, uh, commercial models are implementing watermarking technologies to say, hey, this image or video was created with Claude or whatever model.

That's great, but there's thousands of models out there that don't do that.

So we're dealing with other AI defensive technologies to help identify, you know, images and videos that, um, you know, are not watermarked.

So it's a cat and mouse game.

Um, but yeah.

- But there are things that organizations can do to strengthen their defenses against the AI scams.

- Absolutely.

Yeah.

I know everyone's looking for an easy stop AI button or software or something they can install, but in my experience, working in cyber investigations for quite a while, um, is, uh, training, training of your, your, your population, your employee base, um, to, to notice when something's off implementing multi-factor authentication throughout your organization, having stops built in for, especially for financial transfers or access to higher level sensitive databases where you can implement other levels of verification before access is granted.

- So before we go, I want to get one last thought from each of you on this whole issue of AI.

I know it's a topic that we could talk about forever.

Christian, I'd like to ask you, as the journalist watching this from the perspective of community reporting, what do you think that the public still doesn't understand about AI?

- Yeah, I think a lot of people come to meetings or are against AI developments, having kind of read valid media reports of the worst that these developments have to offer in terms of noise pollution and environmental degradation and what it can do to their community.

And a lot of conjecture on social media is important to remember that every AI development is different.

No two are the same.

It's very much in the context of your own community, what your local water supply is, what your environmental conditions are.

You know, the clearest view you're going to have of something is in your own backyard.

So it's important that if that is coming to your community, paying attention to what our local advocates saying, what our local politicians saying, what is the plan that is public and kind of looking at it that instead of a whole big picture around the country.

- Dr.

Mali, you know, in academia, um, you all have the job of training the next generation of, of, of workers and researchers.

So how do you balance the need for the guardrails that AI experts are calling for that are very alarming, and the need to produce a workforce that can deal with and succeed in a world where AI is part of it.

- So in my view, I don't think that innovation and guardrails are opposite sides of the spectrum.

Right?

They are same.

We can create a responsible AI guardrails, uh, in which your understanding where it should work and if it's sensitive areas such as defense, military, cyber security, we know what are the harms with, with respect to this.

So this is where universities come in.

This is where academia comes in, because we are not teaching them just how to build these technologies, because these young generation should have AI literacy as well.

So how to build them, how to use them ethically safe, responsibly.

And what will happen if you deploy these kind of systems in high risk settings, right?

So you should know what is good.

What is a bad side of this?

This is where AI comes in.

And with respect to guardrails, as we have seen with all the technologies which exist in this world, there's always this, there's good, there's bad.

AI is not no different.

It's similar to what we have seen in ages.

Right?

So with respect to AI, we need to develop a system which is understanding the ethical aspect of what AI is rather than saying that, okay, let's stop the innovation.

Let's have one button which says, okay, let's stop it.

- Well, speaking of buttons, right, Dr.

Ali, is there a kill switch?

Um, in cases where the AI has gone out of control.

- Yeah.

- Based on what you've seen in the healthcare industry.

- Um, you know, the way I look at I'll divide this into two two parts.

The first part.

People think that AI is a crystal ball.

It is not a crystal ball, right?

It is.

It is not a globe which tells you what it is.

Think of it more as a smoke detector.

The smoke detector doesn't tell you where the where the fire is, nor it produces firefighters.

It just hints you where that problem is going to be.

But some kind of a responsible way to hold it when we can.

We even have a kill switch for humans, unfortunately or fortunately, whatever.

However you think about it, you know there is a switch if you have a water leak in your house, you have a stop switch for the water to anything which can go wrong can potentially or will go wrong.

So I think for protecting all of us, there should be some regulatory requirement.

- Sonny, what are some of the things that individuals practical things that we can do to protect ourselves?

- I'd say there's a huge list and we don't have a whole lot of time, but I would direct people to Google Operation Winter Shield for the FBI.

We have a really good list of things people can look at that we launched a campaign last winter and they can go down that list.

And if there's just 1 or 2 things on that list they implement, it will.

It would definitely improve their cybersecurity for them and the country as a whole.

I would see less cases come across my desk, I guarantee you.

So there's a huge list of individual things people can do, but I'd rather point them towards that good resource.

- Because so many of us feel that there's nothing we can do to protect ourselves.

- I will say, like the other people here have said, AI is not an easy button.

You know, it's not a everything all in one.

So it's people.

We are usually the weak link.

So it's us doing our own hygiene, which is the best defense.

- Thank you so much for coming in, all of you.

We appreciate your time and your expertise.

Before we go, as we look at the rapid evolution of artificial intelligence.

A new episode of PBS Independent Lens is also asking a bigger question who is shaping AI and how is that shaping our world?

- Silicon Valley has always liked to pretend that it doesn't have a history.

Some of that history ain't so savory.

- We're back to a little bit of hype.

- The tech bro mindset.

- At its core.

This rise of what some people call techno fascism.

- Big tech don't want to be accountable.

- To show up here today because we're.

- An executive from X AI ducking out a side door.

- They just bad boys misbehaving.

- This is definitely a high IQ group.

- The Ghost In The Machine examines the history of AI and the human values, biases, and power structures behind that technology.

You can watch the new episode of Independent Lens at pbs.org/independentlens.

Again, we want to thank all of our panelists for being here today.

Thank you so much for your expertise.

We know you have plenty of choices for your news and information, and we thank you for choosing us.

[Music]