Tapping AI: The Future of Customer Experience at Verizon Fios

Justin Reilly, head of customer experience innovation for Verizon Fios, is projecting that 80% of value creation in artificial intelligence (AI) will be in business-to-business [B2B] applications, and the rest in consumer services. In an interview with Knowledge at Wharton during the Mack Institute for Innovation Management’s fall conference on Emerging Innovation Leadership Challenges for Global Firms, he also noted that Generation Z is increasingly becoming protective of personal data and may start withholding it from companies unless offered compensation, while millennials will continue to favor companies that make positive social contributions. Reilly, who was a conference speaker, also suggested that in a world of vast automation via AI, companies that learn how to preserve a human touch should have a competitive advantage.

An edited transcript of the conversation appears below.

Knowledge at Wharton: Could give us a synopsis of your conference talk?

Justin Reilly: I spent a lot of time talking about how we think about innovating inside of a big company. I’ve been an entrepreneur my entire life, so this is the first time I’ve worked inside of a large company. I chose a Fortune 14, I guess, to be my first time.

We talked a lot about Generation Z trends, about how we think about becoming a product company, how we shift organizational structures, what it takes in our opinion to keep up with emerging tech, and then how we’re solving some really key problems in our business as we face the reality of being disrupted from a lot of different angles.

Think about the genesis of Verizon, the phone company that is now playing in everything from fiber to media to artificial intelligence. Those things are happening at scale all over the world. So it’s an exciting place to be — there’s a lot of change.

I have responsibility for our core innovation products, for everything that touches the customer outside of our core TV and fiber products.

Knowledge at Wharton: What are some of the most innovative things in that category?

Reilly: I have an entire group focused on artificial intelligence and machine learning. Mostly that manifests itself in a conversational UI [user interface] experience, affectionately known as a chat bot. So probably the most consumer facing thing that is in production that customers are playing with right now is a chat bot experience that we have in Facebook Messenger … and it solves every problem from answering the question of what’s on tonight, getting TV recommendations, to being able to reset your router, to know where your tech is, to pay your bills, etc.

Knowledge at Wharton: Tell us more about AI and machine learning. There’s also a lot of hype around it. It’s like trying to separate fact from hopes and dreams, and a lot of those hopes and dreams will become reality one day. But let’s say for the next two years, which things will actually happen?

“The most interesting trends in AI are going to be in supervised learning.”

Reilly: There are a couple of problems with what is going on with AI right now. People are referring to things that aren’t core machine learning neural networks — what would be referred to affectionately as AI — as [if they were] artificial intelligence: basic linear decision trees, basic data analytics. Certainly [these are] building blocks towards artificial intelligence, but when folks flood the market with “we are an AI company,” or “we are doing AI,” it becomes really hard to differentiate between those things.

Knowledge at Wharton: A simple calculator, in a way, has artificial intelligence.

Reilly:  That’s right. Especially if it remembers something you did. The most interesting trends in AI are going to be in supervised learning, which is a little bit further along, and much further along than unsupervised learning. [It provides] the ability for you to make a couple of interactions with a piece of technology … [and] for that machine to learn what you are doing. But then, it is taught [new skills] by an employee, a product manager or a machine learning engineer on the other side.

For example, we had a chat bot interaction during the Floyd Mayweather, Jr.-Conor McGregor [boxing] fight. Over time, when people were coming into the chat bot to purchase the fight, they would use words that were MMA [mixed martial arts] slang, for lack of a better term. And because we had natural language processing, understanding and generation interaction as well as a neural network behind the bot, with a little bit of supervision from a machine learning engineer, it learned those terms quicker. So when the next person came in they knew that this type of language meant this: Show this result, give this experience.

That, I think, is going to be the most interesting set of interactions for any brand going forward. But it’s fundamental to giving someone an experience where they can give you that type of data. And then the second thing is whether you have the right data, that it’s organized in the right way, and then you can build learning algorithms based off of it.

Most of the benefit in AI probably is going to be in large-scale trucking and warehousing and infrastructure — anomaly detection. Because all of those things don’t necessarily rely on a human interaction, so you don’t have to spend a lot of time thinking about how the things interact, you can focus just on the data set, the learning, the feedback loop. So I would say if you were to try to pie-chart the world  — probably 80% of the value creation in AI is going to be in B2B or large scale infrastructure, and about 20% in customer value creation.

“Probably 80% of the value creation in AI is going to be in B2B or large scale infrastructure, and about 20% in customer value creation.”

In my opinion, the 20% is going to happen in places like customer service where there’s a lot of costs for the business so they’re very incentivized to make the experience better. Customers don’t like, for example, sitting on the phone with someone for two hours. So there’s a lot of a kind of symbiotic relationship between brand investment and customer needs that will allow for an investment strategy. Everywhere from inside of brands to venture capital, if you look at a lot of where the AI money is going from a VC perspective it’s either in large infrastructure or core customer-focused needs like customer-service interactions.

Knowledge at Wharton: You have projects for this year, next year, three years down the road. What about a little bit beyond that?

Reilly: For us, because of our interaction with the home, we’re going with fiber right into your home, we have a tremendous amount of intelligence of what happens on the network … to provide a better experience, to provide more utility.

What I think is going to be key for Verizon, specifically Fios, in the future is how that intelligence gets better as homes become smarter, and how that will improve your utility over time. It’s very easy to see a future where everything in your home is connected, they all speak to each other, they’re running on a fiber network so it’s very clear, secure and fast. And you constantly are leaning into that experience — being better.

There is a lot of thought, though, that you’re just going to have a world of a bunch of personal assistants. I actually think that the personal assistant space is going to be crowded in the next three to five years, and if you imagine the kind of app atrophy, or app kind of clutter that we’ve experienced on smartphones, you’re going to see that same thing with personal assistants, whether it’s Alexa or Siri or 40 other things that you will have to evoke with a wake word to make that thing happen.

And then … about five years out, it will converge back to a couple of key players and we’ll get it right.

Knowledge at Wharton: What about security?

Reilly: When I got to Verizon about two years ago I noticed that we had a long way to go on pure digital product innovation — having our digital experiences feel like Google, Facebook, Amazon. But one thing that we do right is the network.

We’ve been a phone company … we’ve been a network company, for a long time. And we constantly get the network right. If you were to look at where our investment is, we’re spending billions of dollars to make sure that in smart cities, smart homes, we have not only just the network infrastructure to make those things fast and work well for you, but to make sure that they are secure over time.

It’s one of the benefits of fiber — the feedback loop of all of the things going on in your home when you connect to Wi-Fi through fiber. It’s tremendous. We’re constantly looking at the next generation of routers that have deeper security, IoT [internet of things], MQTT [or machines-to-machine IoT connectivity] protocols — all of these things that drive session management and overall semantic security inside of your home, that sequence the sessions, provide cryptography, that provide anomaly detection … so that you can trust us with your data. That is the forefront of everything that we’re doing from an engineering perspective.

The problem with the smart home market right now — and if you look at the trajectory, it’s been relatively flat for the last five to eight years — is that when someone buys their first smart home device, if they don’t get immediate utility out of it they never buy another one. Once they buy three to five, they’re more likely to buy 10, and roughly they’re on a path to buying 20.

Knowledge at Wharton: So what are these 20 things?

Reilly: Most of them are light bulbs, thermostats, security cameras, things that make a single room more useful. So what you see is better adoption if I make a room smart first versus making different parts of my house smart, because the total utility of you as a human in that space gets better. That is the key focus. What is holding up the smart home market is the standard for management across all of these different hardwares. Everyone is trying to figure out what that means.

“There are a whole host of startups thinking about, what does it mean to de-enroll your data from entire brands?”

Knowledge at Wharton: So it’s the old VHS/Beta issue?

Reilly: Right. And you can’t manage them all from one place. There are a bunch of hubs that people were trying to play in. I think our opportunity as we ship more and more smart home intelligent internet hub experiences is to drive light management of all of those things so that you have all of your things in one place. And if you need to go deeper with the usability, you can pop out the Nest app, or Canary’s app and have a deeper experience.

Knowledge at Wharton: What will be those things that will excite customers?

Reilly: If you spend a lot of time with customers in their home, the things that we do every day are the things that take a lot of time, right? Taking the laundry out of the washer and putting it in the dryer, pushing the toaster down, running the dishwasher. I fundamentally believe that as those appliances become more intelligent, a lot of the human utility will come from more intelligent dumb devices that we currently have in our house.

[However,] I’ll give you an example … [of] the problem with a lot of what is being built today. There was a refrigerator that was out about a year ago that would tell you if you were out of milk. So you could ask, well do we need that? And what I would say is, I don’t think we need that.

A refrigerator came out about six months ago that doesn’t do that, but when you tap the door it makes the door translucent. You can see through it and see all of the things in there. That’s a light incremental innovation that is really great. We’ve been thinking a lot about this, and I fundamentally believe that if you make something that is “dumb,” for lack of a better term right now, more intelligent, you have to keep the core of what that thing is.

What do people use refrigerators for today? They leave notes for the family, they hang things that are important, right? They tell stories on their refrigerator. When you spend time at people’s homes, that’s what most families use refrigerators for. So why are we trying to build features that don’t do at least that thing better than it does today?

Maybe the translucent thing is actually the more useful thing. Maybe it’s an amazing, rich storytelling hub for your home that allows you to leave notes for your loved ones, that allows you to interact with other parts of your house. If you take that approach to every single thing in your life, every one of these smart things, it starts to inform how you actually drive value.

Knowledge at Wharton: What else is it that is important to know about where this AI world is going?

Reilly: That’s a large question. I think large brands playing in artificial intelligence have a fundamental responsibility to humans to get this thing right.

What I mean by that is we’ve all gone from talking to each other as humans to looking at our phones as we walk down the street. We’ve all had these things change the way we interact with humanity. The next wave of contextual, personalized, artificial and highly intelligent cognitive experiences have the potential to push us into a world that has much more disassociation with reality, with people, etc.

Knowledge at Wharton: One observation about Facebook has been it considers itself at times to be just a platform.

Reilly: I fundamentally disagree with that if large percentages of the world interact with you as a brand, and get large sets of information [from you] — for example, 61% of Americans get their news from Facebook and they trust the things that their friends post. For large sets of consumers on our network who are doing a lot of things, we have a responsibility as a brand to build great things for them that drive value in their lives. But also, we have a responsibility to make sure that we are doing what is right for the future of how our children interact with technology.

Knowledge at Wharton: What does doing it right look like?

Reilly: There are a lot of things that brands could do that would drive a lot of cost out of the business, drive a lot of revenue into the business, based on personal information, that would drive sets of behaviors from customers, that maybe aren’t the right thing to do.

Given the technology today, and the pathway that we’re on, we could probably fully automate our entire business. There are a lot of brands that will do that…. In the next three to five years, some of this basic AI automation will just be a free, open source tool that you can use.

Just like Squarespace has done with building websites — 15 years ago building a website cost $100,000 — it was a huge thing. Now you can go to Squarespace and get a free one. AI is going to be no different. But I think we have a responsibility to keep the human touch in a lot of these experiences for the better good of both the experience and humanity.

“This next generation of consumers is going to blow a lot of these things up just in fundamentally how they think about the world.”

You could also argue a different angle, which is it’s actually a competitive advantage to have the human touch in a world of full automation. So if you imagine where we’re all walking around not actually talking to each other, and interacting with all of the brands and content and music, and all of these things that we like, but [doing it] completely through whatever device. Maybe it’s just everywhere. Maybe you can just walk up to everything, and they know it’s you.

Then maybe human touch is the differentiator in that world. And maybe it’s also contributing to a containment of disassociation with the world. A lot of futurists talk about this moment in time that could possibly happen where large groups of people unplug. They pull themselves away from the matrix, for lack of a better term. I don’t know if that thing will happen at scale as much as people want it to, but I do think if brands get this thing wrong it definitely will happen.

So how do you do that world? Something just perfectly simple I have done has changed the way I interact with technology. I don’t have any notifications on my phone anymore. I’ve turned all notifications off, because I travel all of the time, and there are thousands of emails. This thing was driving my life. And as I have turned notifications off, I’ve reclaimed my time. I’m the person in tech thinking about this, and I’ve said I’m not going to have these notifications. I have one or two things with emergencies, with family members, etc., that come to my home screen, but everything else I decide [whether it has] to go in.

Knowledge at Wharton: Do you think that companies will compete on this idea?

Reilly: I do. Think Toms Shoes. There was a big period of time when there was a delineation between for-profit companies and not-for-profit companies. And then Tom Shoes came in and said, “Hey, we’re a hybrid. If you buy a shoe from us we’ll [also] give it to [the disadvantaged].” Millennials love that, because there’s a sense of social good in the millennial generation.

If you look at what Generation Z expects from brands, they expect utility in return. Not just a better experience, but utility in return for what they view as currency which is their data. So if you are not very clear on what the utility is that you are driving from them, you’re not even going to get a shot at my data.

There are a whole host of startups thinking about, what does it mean to de-enroll your data from entire brands? What does it mean to say, “Hey Google, you no longer have — or hey Verizon, you no longer have — the rights to my data because you’ve done something that I don’t stand by. And so now I’m going to pull my identity away from your platform.”

There’s nothing to suggest that that won’t become a thing that empowers individual humans. So then in that world, what is your differentiation? … What makes you a better company — and it doesn’t have to be social good. It should be, but it just has to be how you approach the world, and then how your users feel about that.

I’m not sure the next generation is as concerned with privacy as privacy proper as we think about it. Millennials certainly, and every generation before that. I think Generation Z just says, my mobile telephone number, my address, my email address, these things are currency. So if you think you just get the right to those things — I’m already paying you real currency, right? So why do you get this right to this thing? And the experience economy, the social media revolution has kind of resorted us to this idea that if the thing is free, then we pay with our attention. But we haven’t had that conversation around the data component and what that means.

I think this next generation of consumers is going to blow a lot of these things up just in fundamentally how they think about the world. They’re just more advanced, more thoughtful. They’ve never grown up in a time without technology.

I’ve been in co-creations where a seven year old has said, “Hey, why did you put the hamburger menu on the right side? Your shelf nav should do this thing.” And you’re thinking, how do you know these words? You’re not even a trained user experience expert. But this is how they interact with technology, and their knowledge is just far and away superior….

Knowledge at Wharton: You’re talking about a digital consumer boycott that could go viral, right?

Reilly: And maybe it’s table stakes. Maybe it’s a feature that you can just opt into. And you can say it’s a boycott, or you can say at some point it becomes business as usual to how you interact, or how you connect into the world. Your plug in moment is now informed. There are a lot of things like this where people are dabbling, ad blockers, all of these things, but imagine if you actually had control over your identity….

Knowledge at Wharton: It’s so interesting because everything for years has seemed to go the other way. Service provider are just collecting data and we have no control and all of that. But now you’re saying people may wake up and say — or at least the Generation Z may — that this is kind of like my medical records, this is my private stuff. And you don’t get it without me having trust in you, or me liking the use that you’re putting it to.

Reilly: And to that point, if you look at something like blockchain: If I no longer need you — said bank — for you and I in this moment to exchange funds, I don’t need a third party to validate that — because our crypto handshake is good enough for you and me. Then where is the data transfer? You and I don’t need to transfer data, so what happens in that moment?

To your point about electronic medical records, there’s a lot of really interesting work done in this space, because if you look at infectious disease or oncology, one of the biggest problems in care is not being able to take the most perfect information of your visits across different providers. And so if you had identity management that also encompassed everything in your life, you could show up with your full medical history in your identity management software, for lack of a better term … and say, “Hey mister provider, I just pushed this thing to you and now you have perfect information.” I don’t have to wait for the fact that, yes, five years of my medical history is in an EMR, but the other 15 are in stacks of paper all around, and we’ve got to scan it in and fax the document to you, something I know you’re not going to read. So there’s a lot of upside in this idea across industries.