Welcome to season three of MTE's Plugged in podcast.

This season we will continue to explore the world of electric vehicles.

Whether you're a seasoned EV driver or you're just beginning your journey into electric vehicles, this podcast is for you.

For a more enhanced experience, be sure to watch the video version of this podcast that will be linked in the show notes.

Hello, everyone, and welcome to MTE's Plugged in podcast.

Brandon.

I'm Brandon Wagner and as always, I'm here with Amy Byers.

Hey, Amy.

Hey, Brandon.

And today we want to welcome to the podcast John Harris.

John is MTE's Vice President of Information Systems and technology, and we have asked him to come and talk to us a little bit about artificial intelligence, AI and cyber security and kind of what that we're going to talk in general terms and then we're going to pull in the EVs and see what he has to say about that.

So welcome to the podcast, John.

And before we get started, tell us a little bit about yourself and what you do here at mte.

Sure.

Well, thank you for having me.

I've been here at the cooperative for 22 years, started in our, what is now our substation maintenance group, and through various roles here at Middleton Sea Electric and projects that worked on through the years, I found myself as the Vice President of Information systems.

So thank you.

Glad to be here.

Well, great.

I think you have the first question.

Yeah.

So, John, we're going to talk a little bit about AI, but AI has very much become a buzzword.

And so I find that when people talk about AI, they talk about it in different ways.

There's large language models and there's different ways people interface with it.

And can you just before we dive real deep into AI, kind of give us a baseline of what is AI?

Sure.

Well, AI has really gained traction, I would say, in the last four years with the exponential growth that we've seen with it.

Traditionally, AI, I would say, is more machine learning product where you want a specific output or a prediction and you feed it historical data to let the AI make a decision on that and either give you a current state prediction or a future prediction based off of historical data.

More recently, we've seen a lot of advancements in large language models like Copilot, ChatGPT, Grok, whatever your favorite, favorite AI of the day is, where you can basically type it any question or ask it to make any hypothesis, and it will try to mimic the human brain and come out with a reasonable output for that.

Of course, it's still not a human brain.

It's taking in information that it has access to and then making reasonable judgments on that and trying to give you an output based off of the information that it has.

I find that I think I use AI every single day and I feel like I'm.

I heard someone say all the LLMs are different.

Some have access to the Internet, some batch the Internet and reference old their data.

So it depends on what you want to look up.

But I've heard some people say try to use an LLM before you go to Google and I have found that it's been more helpful and it's customized.

The answer is customized to what I need rather than going finding a Google thing and then trying to apply their situation to mine.

But let me ask you, outside of just using LLMs and maybe including using LLMs, what are some ways that MTE is using AI and into that predictive stuff and machine learning and all of that beyond just the LLMs, what are some ways we're using that to make some impacts for our members and maybe some ways that other local power companies are using it.

Before you answer that question, why don't you tell us what an LLM is?

Because I'm not 100% sure I know what AI is.

I know what Jatpt is.

That's my best friend.

But I don't know what you're referring to.

Good question.

And we were saying AI.

We didn't really actually say artificial intelligence.

But LLMs are large language models and there's several that are real popular.

ChatGPT is one.

They're one that batches the Internet.

I think it's great for creative and trying to come up with Give me some ideas.

ChatGPT is great.

Grok actually accesses the Internet so I use it more as a Google replacement at times.

But it'll still operate like ChatGPT and Grok is part of Xai.

That's the Elon Musk flavor of AI.

There's one called Claude that I think is really good at solving puzzles.

And so when you need real deep thinking, there's a ton of them.

Deep seek.

Jessica told me the other day about hugging face, which is a place where if you're building, you can actually share what you're building.

And so it's just amazing.

It just continues to grow and grow and grow.

So ask your circle back to the question.

I don't even remember what your question was to John.

So I mean, obviously we can use LLMs to help us with some of the tasks and Becoming efficient at what we do.

But what are some unique ways that local power companies can use AI?

So there's the probably the obvious things that a lot of us use AI for, whether it's to helping craft an email.

We use copilot in meetings to let it do a translation of what is said in the meeting and then create a summary or, or a minutes of that meeting afterwards.

So those are I would say some of the generic copilot type things that we're using it for.

But there's any number of industry specific things.

For instance, our members can look at their usage data broken down by appliance through an AI product called Bijli that does a load disaggregation.

You know our data only has how much kwh you use for a given 15 minute interval or whatever the case may be that digitally AI product can look at that 15 minute interval and make assumptions based on time of year, weather patterns, things of that nature and make a pretty good guess about your house, how many square footage you have and make assumptions about how much use is your air conditioner, all your major appliances, your air conditioner, your water heater, refrigeration, lights, things of that nature.

So that's a member facing thing more internally.

They're getting away from large language models into machine learning.

There's predictive analysis that you can that utilities are doing with their load to understand where their peaks are going to happen so they can better manage their load to reduce their demand bill from their wholesalers.

And we do that today.

There's gosh, there's any number of things taking images of your system and using that for maintenance purposes to find bad poles, broken insulators.

Also taking those pictures and looking at your right of ways.

Some are using satellite imagery to detect trees that are dead or dying.

What type of tree is it, how fast do they grow so that they go cut where they need to instead of doing a complete clear cut cycle like we do today.

So yeah there's and there's plenty of use cases out there and they're just going to continue to grow.

System planning I would say would be a big use case for it in the near term.

So you mentioned I want to circle back a little bit to something you said you mentioned.

Begley and I do want to make sure that all our members know that they all have access to bitchly.

You know we talk about it on what we can learn from it and what the members can learn from it.

But you don't have to sign up for this or anything.

You already have it.

If you have the MyMTE app, it's part of that, it's integrated with, into that app.

So if you look at your usage within that app, you're going to see that it is also what is sending you an email in the middle of the billing cycle.

It sends out an email and says, hey, we're projecting.

If you keep up what you're doing, your bill is going to be this.

And it's a great opportunity because I know, you know, during the really cold spell, we would get that email and we were like, whoo, we need to slam on the brakes.

And we would, we would really, for that last half of the month, we would really try to kind of change what we're doing, be a little bit more conscious of the energy we're using.

And then the weather may or may not have played out the way AI thought it would.

So maybe the temperatures were not quite as bad, but when we got our bill, it was significantly lower than what it was predicted.

So you never know.

But it is a great tool that I think we have for our members to.

I always say what we're trying to do with the MyMTE app and with this Bijley is we're trying to give our members control over their electric bill, give them all the tools they need to make decisions on their own to conserve, to save money, those type of things.

And I think Begley has been a great tool in that toolbox for them to do that.

I agree.

Yeah.

And those cold, really extreme cold months that we had where maybe we have a really, a steep dip in the temperature.

Yeah.

It can maybe over predict what your bill is going to be, but again, I think that'll just continue to get better over time.

I think in the more moderate months, it's pretty spot on, probably.

Yeah.

I know we had a very cold winter, I think, and we had a lot of extremes, like it would be really, really cold and then 80 degrees.

So, I mean, it was a crazy kind of winter, so it probably was wreaking havoc.

But hopefully Begley learned a lot with Flora.

I still can't predict the weather much better than anybody else can.

No, so true.

So true.

So true.

So.

So beyond the energy world, there are a lot of emerging trends in AI.

Kind of want to pick your brain on that.

And then I also want you to talk to us a little bit about tying in this AI.

We're all kind of using it, but how is that when you're thinking of cybersecurity?

I mean, is it learning from you to where it could act like.

I mean, there's the good side of AI.

We're all loving it.

There's the bad side where it's robbing your voice or whatever and calling your parents, saying, give me money.

So can you talk a little bit about that?

So with emerging trends, every software product that comes out seems to have its own flavor of AI anymore.

So every software vendor out there is trying to figure out how to capitalize on the AI market with cybersecurity.

That's really a dynamic of one where our cybersecurity security vendors use AI.

We have a ton of logs that an AI engine can parse through and look for anomalies and attacks or really anything that's not.

That's a change in our environment.

Right.

And they can detect that.

The other issue with that is, like you were alluding to, is the attackers, the bad actors are utilizing AI as well.

And can our vendors, our good guys, keep up with what the bad actors are doing and.

And stay on top of that?

It's kind of a scary world in that regard, I would say.

But I feel like overall the good will prevail, hopefully in that realm.

I hope so.

I know, you know, people that, you know, are elderly and people that maybe are not in the business world right now dealing with eight I, you know, they may not understand that, you know, you get these phone calls or you'll see something on the Internet, and it was like, well, they had a picture and it looked real.

I'm like, that was AI.

That's not real.

You know, so it's really, I think, falls to us to really look after our parents, our grandparents, our friends that may not, you know, they've retired since the AI has come on the scene.

But it falls to our, you know, us to have the responsibility to look out for them.

I feel like, absolutely.

If there's any way you can have influence with someone in that situation to try to help them understand that if they get a call that doesn't sound right, it may not be right.

Or if they get an email from a known entity or what they think is a known entity, it may not be real.

And they're getting better at fooling us all the time.

They definitely are.

So now I think we need to shift and talk about EVs.

This is an EV podcast.

Well, you know, so, John, those of us who are driving EVs, I feel like we maybe get a real dose of AI in real time, because I know even my car, they've changed the way they do autopilot using end to end neural network training.

They've started telling me we're going to be able to predict your mileage usage better because they're using AI to see my patterns and driving patterns.

But what about the cybersecurity part of that?

I know that it almost feels like we need to have a healthy amount of cynicism with things.

If something sounds too good to be true or sounds so bad that maybe be cynical about that.

But as an EV driver, public charging all these things, what advice would you have for, for us to keep cybersecurity in mind?

One thing I would say is make sure your software is always up to date.

Any known attacks or any vulnerabilities that your vendors are gonna be aware of and they'll patch it.

And the more you can, just like with your phone, the more you can keep it up to date with whatever version you have of software.

That's probably going to be your best protection.

But really being cautious, being aware of your surroundings is always important as well.

I saw a quick video this weekend of Justin Meyerhofer with TVA who said I thought was really interesting when you think about cybersecurity.

He said national security is the same as energy security or vice versa.

We need to think of it in those terms.

If our energy sources are not secure, then that has an impact on national security.

So you know, obviously two EVs are a part of that.

It's maybe a mobile part of the energy.

But what are some ways that you all.

I know we don't generate energy, so we don't have to worry about someone hacking into a generator and shutting us down.

But what are some ways that you're making sure, you know, you got your business network, you got your industrial network, and how does that maintain without letting in the bad guys you mentioned?

So our operational network, which is where our scada, our devices, all that lives, we.

There's a term called air gapping, the network where it doesn't have access to our enterprise network as much as possible and certainly it doesn't have access to the Internet.

That's one method.

Other things we do are making sure that we're using encryption protocols on all of our data in transit.

So communicating to devices from our SCADA environment, making sure that we've got good solid encryption methods on those items, but mainly is keeping that network isolated and the correct firewalls in place so that bad actors can't get into that network.

The other side is we still connect computers to that network as well and we've got to keep those up to date and make sure that we don't have any unauthorized computers getting to that network.

One of the things that my co worker actually told me about today is we get a lot of.

You test us, you all test us quite a bit to make sure that we're not clicking on anything we shouldn't click on.

So that's part of the training too, right?

Training the humans to make sure that they're.

Yes.

And studies I've looked at recently, still, the end user is the biggest risk that we have.

That's me, that's you guys.

Right?

And I would say even in the technical field, you may even get attacked even more.

So you have to be diligent, make sure you know what you're clicking on, because it really just takes one click and it could, you know, hopefully we would detect that fairly quickly and you may end up getting your computer wiped out and starting over.

And you think you, you know, you think it's going to be obvious, you know, that you're like, I would never fall.

I would never fall for that.

And I'm usually pretty good.

But I have to say, y'all got me a few months ago and I was like, so disappointed in myself.

But you sent one from, like, it comes from however you do it.

But it was talking about Microsoft Planner or something.

And the thing was, we had just had a meeting where we were talking about Planner.

So I was like, well, this is probably tied into that.

And I was like, who has assigned me something?

And I clicked on it and I was like, oh, my gosh, they got me.

I was so mad.

But you get caught up in your work or whatever and you get these emails and.

And you're probably not really focusing like you should.

I mean, I'm always so proud of myself because I'm like, well, they never get me.

And then I was like, oh, my gosh, no.

We all get a high volume of emails and it is hard to stay focused and diligent to not click on the wrong email.

Yeah, yeah.

Luckily it was you testing and not someone trying to break in.

So.

So before we let you go and kind of looking ahead, I want to bring it back to EVs.

And we have talked about autonomous driving, We've had podcasts on that, and that's all about AI driving your vehicle.

And I have said many times on this podcast that I've seen too many movies, too many sci fi movies where that goes wrong.

And I would never get in an autonomous driving vehicle because they always are Going to drive you off a ledge or they're going to take you over.

You know, I've seen a lot of that.

So with that in mind, what kind of, you know, cyber security challenges do you think, when you're talking about autonomous driving and the interconnection of all these EVs and everything, what do you think about cyber security with that?

I mean, is that going to be safe or would you get into.

There's the question, would you get into one of those vehicles?

I probably would not.

Are they safe?

I can't answer that.

I don't have a lot of experience with them.

But from a cyber security hacking standpoint, I would say that is a very vulnerable thing you would want to consider if you are doing that or have plans to start looking at products like that.

I mean, those advancements are coming, it's just a matter of time.

But yeah, the cyber security aspects of that would certainly give me pause.

Yeah, I think so too.

And I've seen, I can't remember, it might have been Mississippi State, I can't remember, but they had like electric buses that were autonomous driving that are taking the kids from one end of campus to the other end of campus.

I think they're kind of open air.

So I feel like you could jump off if things went south.

But I don't know.

But I was like, wow, it's, you know, it's here.

You know, we've talked about the robo taxi and that kind of thing, but I just don't know.

I don't have the warm and fuzzies about that.

It is here.

I saw a video again this weekend of Tesla.

They're using unsupervised full self driving at the end of their production process now.

So when the cars are done off the production plant, it drives to where it's supposed to go and park.

They've logged 50,000 miles in those conditions.

So yeah, it's, it's.

I don't think people realize how close it really is.

Well, I think I'm just going to stick with using AI to help me write emails.

I don't want them driving my car.

But John, thank you so much for being here today.

I think this was very, very interesting.

Very good topic, very timely topic, very informative.

So thanks to everybody who was listening today.

For more information on MTE's Drive EV programs or the VVCAR Club, go to DriveEV.com or email us@evcarclubte.com and until next time, plug in, power up and drive safe.