Taste of Informatics

Deep Learning & Artificial Intelligence: Opportunities across business and society

April 27, 2021 Informatics+, College of Informatics, Northern Kentucky University Season 1 Episode 3
Taste of Informatics
Deep Learning & Artificial Intelligence: Opportunities across business and society
Show Notes Transcript

The opportunities are not just restricted to those that build AI, but the future also belongs to those individuals that can employ it.

Host Mike Nitardy and guest Todd James from Fidelity Investments have a wide-ranging conversation about the transformative nature of artificial intelligence, discussing what AI is and what it isn’t, and how AI is changing the world around us.

Artificial intelligence, AI, deep learning, prediction, computer vision, machine learning

Brian Jaynes:

Greetings, Northern Kentucky University's College of Informatics, and it's outreach arm Informatics+, would like to welcome you to the Taste of Informatics podcast series.

Mike Nitardy:

Welcome to the Informatics Cafe. My name is Mike Nitardy, and I will be your host. Today's special is on artificial intelligence and deep learning. Thank you for joining us. With us in the cafe today to discuss artificial intelligence and deep learning is Todd James. Todd is the Senior Vice President at Fidelity Investments. Todd, thank you so much for being with us here today.

Todd James:

Mike, thank you for having me. I appreciate the opportunity to talk to you and be part of this event with the College of Informatics.

Mike Nitardy:

Yes, you bet. Thank you. So let's just jump right in. And I guess I'll ask you on behalf of all of our guests, what is artificial intelligence?

Todd James:

You know, that's that's a big question. And I think there's a lot of ways to answer it. But I'll start by saying, at its core, we're talking about a set of analytical approaches and technologies that, you know, seek to address tasks that normally require human intelligence. One of the things you'll hear about often is deep learning.

Mike Nitardy:

Right

Todd James:

Deep Learning is really a subset of AI. It's highly specialized self learning, complex techniques that use a lot of data and a lot of compute power. But it provides us with a pretty powerful way to be able to make predictions. Whenever you think of computer vision, when you think of unstructured text a lot of those solutions are based on deep learning.

Mike Nitardy:

I'm sure that a lot of our guests are probably just like me that when they think of artificial intelligence, they probably think back to the movie with Hal 2000 was it 2001 the Space Odyssey?

Todd James:

The first one is 2001.

Mike Nitardy:

Yeah. And so that being artificial intelligence, or the more recent movies of Iron Man, whatever that thing that he's got in his system that speaks to him all the time, I forget its name. But but it's more than that. That's not just artificial intelligence. Well, I would argue that we don't have artificial intelligence like that now. Thankfully, I mean, you look at the shows, didn't end so well. For Hal and... Right, right.

Todd James:

...2001 Space Odyssey. But I, you know, it's funny, I actually think a lot of people because of Hollywood...

Mike Nitardy:

Right.

Todd James:

...there's a mythical aura around artificial intelligence. And what I like to do when I talk to people about what it is, is really break it down to a to a very, its very simple components. I think the first thing you need to know about artificial intelligence is AI is really seeking to augment or replace human judgment with machine prediction.

Mike Nitardy:

Okay.

Todd James:

So you know, to make it even simpler. AI is predicting a number or a membership in a group.

Mike Nitardy:

Hmm.

Todd James:

And so what that allows us to do is convert judgment problems into prediction problems. And, and when you think in those terms, that opens up a whole new set of capabilities. Like, if you have someone that's processing a transaction, typically they're making a judgment around "is this acceptable or not? Do I do step a, or do step B?"

Mike Nitardy:

Right, right.

Todd James:

But now what we're really looking at is, is making that as a prediction problem, it's more likely to be step two or step B. And there's a bit of a shift there. I think the other thing too, that I like to frame it is when you think of the computer revolution, what the computer revolution really did was commoditize arithmetic, math. What AI is doing is commoditizing prediction.

Mike Nitardy:

Wow.

Todd James:

We're able to do it quicker. We're able to do it at scale.

Mike Nitardy:

Wow.

Todd James:

And I think that's the difference. And that's really a transformative. But as you keep this in mind and when people start to learn about "how do I apply it?" I'm taking a judgment problem, and I'm converting it to our prediction problem. That's what we're doing.

Mike Nitardy:

That's that is really, really intriguing. Thank you. I've never heard it said that way. But that it just that's really just opened it up for me. So what should our guests know about artificial intelligence?

Todd James:

You peel away the mystique. And what you see coming out of Hollywood and the fact that you have artificial intelligence and robots that think and act like humans. That's not what it is. And like I said, Just remember, it's about making a prediction. I am identifying a number or a membership in a group. And I think it's important that people start to think of their business problems you know from a business perspective, or from a research perspective, in terms of, can I convert it to a prediction problem? And I think as they start to do that, they're going to be able to start to identify better use cases. Now there are other components. Do I have the data? Is it something that, you know, have the computer in the processing power to be able to solve in time. But if you can start with, "can I convert the situation I'm looking at into a prediction problem," you're off to a good start in terms of identifying some use cases where you can apply these technologies in the real world.

Mike Nitardy:

Wow, that is, that is something. Now we do have in the title, and I mentioned it. And you also mentioned it a little bit deep learning. And you said, that's a subset? So can you can you get into that just a little bit?

Todd James:

Yeah. And I'll talk first from a usage. We can get into the, the the capability but, you know, artificial intelligence, I mean, can can start with something as simple as like linear regression tied to a straight prediction. But what what starts to happen when you get into machine learning, and then a subset of that deep learning, what you're really trying to do is have algorithms that are self learning...

Mike Nitardy:

Okay.

Todd James:

...and identify patterns and data...

Mike Nitardy:

Okay.

Todd James:

...and draw conclusions. So a real good example of where deep learning can come into play now, is you and I can sit down and read a book. One of the pages can be sideways, it can have good print or bad print, but our minds were able to figure it out and read it. Computers, you know that that's a harder problem to solve. But that's where deep learning comes in for like looking at unstructured text, things like books, being able to look at patterns across large volumes of documents, and be able to identify phrases, words that can be extracted to be able to pull out content to infer meaning. That's where you really start to get into some of the power of deep learning.

Mike Nitardy:

Okay.

Todd James:

A lot of our computer vision is also relying on that. So some of the more advanced approaches, where you're running large volumes of images, or large volumes of documents, through an algorithm that's identifying and trying to select patterns from that, to drive to the outcomes that you're looking for.

Mike Nitardy:

Wow. It's very exciting stuff.

Todd James:

That's high level, there's a lot more behind it. But hopefully, that gives a sense of deep learning.

Mike Nitardy:

Definitely, thank you. And we've, I think that this entire area is cool, but that one of the questions that I always ask our guests is, what is so cool about this area? And how does it help the world? I feel like it's, that's a silly

Todd James:

It's a good it's a good question. I think, I guess question. one. Cuz there's a lot of I think, to some extent, there's fear, what's AI going going to do? But, you know, as I talked about earlier, let's start by saying what are human beings really good at? Right. You know, we're curious, we're creative, we're emotional. Those are traits that are uniquely human. But in spite of that, and you think of your own day, we spend a lot of our time doing manual tasks. We do a lot of simple judgment decisions associated with those tasks. And the beautiful thing about artificial intelligence is it's really well suited to help us with those tasks. And what's the result of that? It frees us up to do those things that human beings are best at doing those things that we like about our lives and we like about our job. Think about when you go home tonight, I mean, artificial intelligence, when you turn on and look at whatever streaming service you have, is going to make a recommendation on what movie you should watch tonight, based on your patterns, based on your history, and what it knows about you. And it's not always right. But it's a lot better than going through an alphabetical list of movies. And we've kind of come to terms with the fact that, you know, we expect it now. And it's a better experience. So you think in the work environment, you have an individual that is processing customer transactions, they have to, they have to look at the transaction that comes in, they have to read it, they have to apply it against a rule set, and then make a decision about step A, B or C. It's not necessarily what gets a lot of those individuals excited about going to work. But if you could start to aid that process, and make it quicker for them, they have more time to interact with their customers or be creative on another aspect of their work. So it drives up job satisfaction, also making them more effective. So those are some of the examples we say every day. You asked about the world. I always like to point out this example. One of the last...I spent a lot of time in India. And one last time I was in Bangalore, I always try and go out and meet some of the emerging companies there. It's a pretty rich tech environment. And I ran across a startup and the problem that they were looking at is according to the W.H.O. during the lifetime, one in 12 women will develop a breast anomaly. In India, the mortality rate is 50%.

Mike Nitardy:

Oh my goodness.

Todd James:

So a little bit different. And part of that's... I was not aware of that. I mean that's access to care. There's some cultural considerations. There's transportation concerns. But the solution that they were looking at was a combination of low cost thermal imaging, where they get a bunch of images. So going back to your deep learning example...

Mike Nitardy:

Right, right.

Todd James:

...and artificial intelligence, and they were they were training data sets to be able to identify heat signatures that result in increased likelihood of a breast anomaly.

Mike Nitardy:

Wow.

Todd James:

And think about this, you take it out in the rural areas, it's non invasive, it's low cost, and you're able to screen a large population. And then when you identify someone at higher risk, then you take them into the hospital in the city, and then you get them care. And, you know, I think when you see solutions like that, that are creative, have a direct line into preservation of life or the improved quality of life. I think we're just starting in terms of the true transformative power of AI. And I think, I think it's getting exciting. So that's always one of my favorite examples of creativity for for, you know, the the betterment of human human society and human life. And you reference this, I think, in your answer, this hesitancy perhaps on certain people that fear of AI, and I'm going to make a bad analogy, but I'm going to try to get your feedback on it. To me, it's a little bit like the baseball debate between sabermetrics and the old way of doing things. You know because I think you said earlier about being a really good at predicting what what do you need predictions for and, you know, sabermetrics took over baseball, it seems like that to where people don't like to steal bases anymore, because it's not worth it. You know, I mean, can you maybe comment on how there's a way to, you know, that will get the benefit out of maybe sabermetrics, but not the downside of no more stolen bases? You know, if that makes sense.

Mike Nitardy:

(laughter)

Todd James:

I think a lot of it too, is is education and engagement. Whenever you have a change, naturally, all human beings, no matter where you sit in society, there's a reticence for change, right? We get comfortable with, with, you know, the the current order of the world. But, you know, what I'm seeing when we're exposing people if you're bringing them in, in the process...

Mike Nitardy:

Rght.

Todd James:

And they're part of the development. And they're understanding the applications. And you're working with them to build the skills that needed to be built for for the new, you know, the the the new order of things. People see opportunity. And I think it's, you know, I think it's up to companies to think that way.

Mike Nitardy:

Right. Right.

Todd James:

I also think it's very important to bring the individuals along. By the way, as you bring these individuals along, you get better AI, you get better technical solutions, because they're, they're weighing in and telling you what they think is right and wrong about the solution and the way it works. So I do think it's a win win for the organization, but also for the employees.

Mike Nitardy:

So what types of jobs would be available in the artificial intelligence field?

Todd James:

All. No one likes that answer, but I'll get to it right? All jobs.

Mike Nitardy:

(laughter)

Todd James:

You know, of course, and I'm sitting here in the College of informatics. We need individuals to manufacture AI. Create it.

Mike Nitardy:

Right.

Todd James:

So we're going to need data scientists, data engineers, cloud architects, ML engineers, all the technical resources needed to build this. But when I say all, we also need product managers who know how to employ the predictive power of artificial intelligence into their solutions. We need operations managers that understand how to work in an environment where you have autonomous and semi autonomous processes.

Mike Nitardy:

Right.

Todd James:

Client Services, representatives, who are able to be assisted by AI guided interactions with customers. So I do think, you know, everyone needs to figure out how to apply it. And so I think raising the awareness of what AI is, what it isn't, how to identify a good use case, and how to work with it is going to be very important, because the opportunities here are not just restricted to those that build the AI, but the future also belongs to those individuals that can employ it.

Mike Nitardy:

Right. Right. Exactly.

Todd James:

And like I said, even lawyers will be using this, you know, something close to home lawyers will be using this to to guide them through review of cases. There's a lot of when you look at document extraction and reading information, and...

Mike Nitardy:

Yes.

Todd James:

...all of our fields are going to have an opportunity to be aided by artificial intelligence.

Mike Nitardy:

No, I agree. That's certainly something I know that that a lot of people in the legal field are looking at. As they should be, as they should be. Well, Todd, I want to thank you so much for joining us in the cafe today. This has been an outstanding conversation. It's just been very informative and educational, and I know that our guests have got to feel the exact same way. Thank you so much.