Used to its full potential, artificial intelligence (AI) can assist employees, improve interactions with customers, and increase efficiency. Rob and Jackie sat down with Nitin Mittal, a principal with Deloitte Consulting, to discuss how AI is being used to enhance work environments.

Mentioned

• Thomas H. Davenport and Nitin Mittal. All-in On AI: How Smart Companies Win Big with Artificial Intelligence, (Harvard Business Review Press, 2023).
• “State of AI in the Enterprise, 5^th edition report,” (Deloitte United States, 2022).

Related

• Patrick Grady and Daniel Castro, “Tech Panics, Generative AI, and the Need for Regulatory Caution,” (ITIF, May 2023).

Auto-Transcript

Rob Atkinson: Welcome to Innovation Files. I'm Rob Atkinson, founder and president of the Information Technology and Innovation Foundation.

Jackie Whisman: And I'm Jackie Whisman. I head development at ITIF, which I'm proud to say is the world's top-ranked think tank for science and technology policy.

Rob Atkinson: This podcast is about the kinds of issues we cover at ITIF, from the broad economics of innovation, to specific policy and regulatory questions about new technologies. If you're into this, which hopefully you are because you're listening, please be sure to subscribe and be sure to rate us. It really does help us get broader reach.

So today, we're going to talk about AI. I don't even need to say what AI is, but I will, artificial intelligence.

Jackie Whisman: Our guest is Nitin Mittal, a principal with Deloitte Consulting. He's the firm's US AI strategic growth offering consulting leader, and a co-author with Thomas Davenport of All-in On AI: How Smart Companies Win Big with Artificial Intelligence. Welcome, we're happy to have you.

Nitin Mittal: Thank you. Thank you for having me on the show.

Jackie Whisman: I wanted to ask you about the research that went into your book. Why did you start it, and what did you find?

Nitin Mittal: I have been practicing in the area of data analytics and AI for a good part of a decade now, and what we were finding is that there's a lot of promise initially around data modernization, advanced analytics, and now more so from an AI perspective. And by working with a number of enterprises around the world, we have been able to help them, we have been able to advise them, we have been able to implement systems for them, but we were not able to find any way of having codified that knowledge, that experience, and the lessons learned, so that it could help the society at large, it could help any business out there, and it could particularly help business managers. That was the genesis and the thinking behind why Tom Davenport and I wanted to co-author and write this book, but there was a lot of research that went into it.

Deloitte has actually been conducting research that is called State of AI in the Enterprise, and that research essentially covers the sentiment, as well as the adoption curve, of artificial intelligence in particular, across many different countries, many different industries, and certainly a vast swath of enterprises that are out there. So we used a lot from that research, augmented by interviews that we actually did with some of the senior executives of the enterprises mentioned in the book, and that together was the foundation for much of what we have articulated in the book and the subject that we have explored.

Rob Atkinson: Yeah, that's great. We've been focused on AI, geez, probably eight, nine years, obviously more from a policy perspective and what AI can do for societies and companies and organizations, and I know that we've relied on that Deloitte dataset you've talked about, they've been been quite helpful to us. But let me step back, particularly for folks who maybe aren't as deep into this, why should companies, and obviously not just what we would call tech companies or IT and internet companies, why should they take advantage of AI and their operations? And obviously, industries vary quite dramatically, but overall, why should a typical mid-sized to large company really be thinking about AI, what do they get from it?

Nitin Mittal: What we've explored in the book, and what we also find through many of the conversations and consulting that we do, is that we perhaps need to think of AI less as a data science endeavor or a technological implementation, but rather, if we take a step back, the way to perhaps think, adopt, and ultimately progress with AI, is a means to augment your employees, a means to augment your interactions with customers, a means to augment your work environment, and the means to augment how you could essentially thrive in the digital economy. AI, at the end of the day, is all about mimicking human cognitive function, the way that we sense the world, the way that we converse, the way that we interpret, the way that we infer, the way that we judge, and the way that we reason. AI is slowly and steadily actually mimicking all those cognitive functions.

Generative AI, which is a lot in the news nowadays, is taking it a step further. So for many of the enterprises out there, the way to think about AI is not about a model or a technology that is being implemented, it's more about, how do I create differentiation? What could be the means and basis for competing in the industry that I want to grow my market share in? How do I be a leader in the digital economy? And what could be the form and function of digital interactions with my customers? It becomes an aid, and it augments all those aspects of a particular company. When organizations think along those lines and with that lens, we see a lot of adoption of AI, and we see AI being embedded in practically every workflow, every process, every task, every interaction. If, on the other hand, AI is seen as just a data science project or a model to be implemented, at best, we see organizations experimenting and conducting POCs, but not have it necessarily be scaled up, and consequently harnessing the promise and potential of AI.

Rob Atkinson: So you start with your processes rather than starting with AI, is what you're saying?

Nitin Mittal: Absolutely. Start with processes, start with interactions, start with tasks, and start with the experience that you want to create, not what you could be doing with a model.

Rob Atkinson: Interesting.

Jackie Whisman: What are some of the more impactful ways companies have used AI?

Nitin Mittal: It absolutely depends, from a industry-by-industry basis. You take the life sciences biopharmaceutical industry right now, one of the big areas of focus with respect to AI is accelerating drug discovery. We just came out of the COVID-19 pandemic, and the fact that we were able to accelerate the clinical trials, we were able to accelerate the regulatory approval, and we were able to accelerate the COVID-19 vaccine that came out, was not necessarily by happenstance. Yes, there was a sense of urgency because of the pandemic, but it was aided and abetted by the usage of artificial intelligence. That has now set an expectation in society. We as patients probably don't have the patience to wait a decade or longer for new therapies to come out. We want that to be accelerated, and that's what the R&D groups in the biopharmaceutical industry are working on, which is how best to effectively leverage AI to accelerate drug discovery. That's one application.

Another application is in the manufacturing space, and the advent of smart factories, the fact that you could actually have vision AI-based cameras, AKA smart cameras, across your factory floor, not only for the purposes of enabling worker safety, but also for continuous monitoring so that you could make adjustments to the production line on the fly. You also have more and more sophisticated manufacturing robots, many of them embedded with AI models that are able to optimize the actual production processes and optimize the ability to manufacture different types of goods.

You take a look at the services industry, and take into account, for example, the digitization of contact centers or call centers. What does digitization of a contact center or a call center actually mean? It is about a virtual digital assistant, no different than Alexa, as an example, that is able to converse in human-like speech with the actual customer. We as customers are not perhaps pressing one, two, three, et cetera, that we used to do in the era when IVR systems were implemented. Now, we are actually conversing with a virtual assistant. But what is that virtual assistant? The virtual assistant is natural language processing and natural language generation conversational AI models.

So that's the set of examples in just a sliver of the economy, but it all goes back to the point that Rob made. Let's focus on the process, the interaction, the experience, the task, and figure out how to embed, how to apply, and how to harness the promise of AI in that context.

Rob Atkinson: It's interesting you mentioned robotics. We had an event several weeks ago on the role of what you could call police tech, in other words, how police forces around the country are using advanced technology. And one of the guests at the event was Spot the robot dog from Boston Robotics, or Boston Dynamics I guess it is. And it's a really cool robot, actually it came by and visited my office and everything. But it's an AI-enabled robot, it can do certain things, and without AI, it couldn't do it.

And then, you mentioned customer service. I was trying to close an account today, I won't say what company, but it's something, one of these financial things, and I've never used it, and I got some weird thing, "Somebody wants..." I was like, "Fine, I'm just going to close it." But I couldn't do it for some reason, so I went online and I did the chat thing, and it was really, really sophisticated. I could have spoke it in, but I typed it in, and it got me to what I needed to be able to do, and I was like, "Wow, five years ago, you would've done this, 'Press one for this.'" So it's gotten a lot better.

Nitin Mittal: Yeah, and that's what we see. And frankly, that goes a long way in terms of this question, what's the adoption rate? How are organizations proceeding? Where are they along the maturity curve? And it literally comes down to, if the starting point is what model to build and way to deploy it, the adoption curve is very low, but rather if it is about, what process do I want to optimize? How do I enhance productivity? How could I change the customer experience? And what is the form of digital interaction that would foster greater loyalty? That essentially helps determine how best to apply AI, where to apply AI, in what capacity to apply AI, and there's a far greater adoption of it, with, I would say, less of a change management challenge.

Rob Atkinson: Yeah, that's interesting, because you mentioned a key point here. This is not just about bottom line improvement, it's also about top line improvement. It's striking, in the customer facing businesses, making electric batteries for Ford is not a customer facing business, but hotels are, and banking, and all that, it's striking how little they know about you in a good way. And increasingly, [inaudible 00:12:42], okay, when I travel, I like to be on a top floor because quieter, I ideally like to be at the end of the hall, away from the elevator, I like a king-size bed, because I'm six foot seven. And the hotels that I go to, the chains, they're pretty good at that, but they're not quite where they should be. Are you seeing that companies are using that also for top line improvement?

Nitin Mittal: We are certainly seeing applications of AI for market growth, top line improvement, as well as creating that differentiation vis-a-vis your competitors so that you could ultimately gain, so we are seeing that. But you do raise a very good point. A lot of the applications of AI to date have been in what I would call the back office, improve productivity, improve a particular process, generate insights that could help with business decision-making, or simulate and consequently predict the various scenarios. I would call that predominantly in the back office.

Where the adoption curve has been a bit lower, for sure, has been in, quote, unquote, "the front office," particularly when you're actually interacting with a customer, where you're interacting with a human being, whether it happens to be in the hospitality industry, whether it happens to be in the fast food industry, whether it happens to be in the customer service aspects of the economy, et cetera. And my perspective on it, and what my observations have been, it is not to do with the lack of sophisticated AI models or the lack of technology, it perhaps has to do with a little bit of innate human fear factor, and the fact that they would be, quote, unquote, "an intelligent machine" in the form of artificial intelligence that is interacting with you, as opposed to the social comfort that we have with respect to interacting with another human being. Maybe it's a little bit biased through Hollywood movies, and there's a sense of, quote, unquote, "creepiness" around it, but that fear certainly exists.

But as we as a society are starting to slowly and steadily overcome that fear, and AI becomes the norm, and it becomes mainstream as part of how we naturally interact in any case... Example, talking to a smartphone, 15 years back, I don't think we would be finding any person walking on the street talking to their smartphone, and if they tried doing it, we probably may give them some kind of a label that...

Rob Atkinson: Who is this crazy person?

Nitin Mittal: Exactly. But now, it's common practice. So with generation and with time, we see that what starts off in the back office permeates into the front office, becomes part of our everyday lives, and when it becomes part of our everyday life, the adoption curve dramatically increases, and we have the mainstreaming of every technological revolution that takes place.

Rob Atkinson: Yeah, I remember, geez, a long time ago, 20 years ago or so when I was working for the Congressional Office of Technology Assessment, and we were doing a study on technology in the services sector, and one of the interesting factors was the delay in adopting or using automatic teller machines, and it was largely generational. Older people were uncomfortable or less comfortable, and younger people were more comfortable. Now, everybody uses an ATM. It's like, "Oh yeah, of course I'm going to use an ATM." So you've got to figure, exactly like that, my 16-year-old talks to Google and Siri like it's nothing, it's just having a conversation, whereas to me, it's still a little bit like, "Oh, this is cool. This is interesting."

Jackie Whisman: I bet your 16-year-old isn't getting cash out of an ATM, though. I bet she's Venmo-ing everything and using Apple Pay.

Rob Atkinson: She would say, "What's an ATM?"

Jackie Whisman: Exactly.

Nitin Mittal: Yeah. We perhaps see these generational shifts, and it's probably best captured by this phrase that there tends to be, I think, in every generation an iPhone moment. Smartphones existed before the iPhone, but it was an iPhone moment that took place in 2007, and that changed the trajectory of essentially, let's say, how we interact with each other, and how we organize our lives, and how we interact with the macroenvironment, where the smartphone is perhaps an appendage to our body itself at this point of time and is very ubiquitous. The same perhaps is happening with generative AI. Generative AI did not, let's say, start the AI revolution, it has perhaps given wings to the AI revolution, and it is its own iPhone moment, where AI has now gone from the realm of the researcher and the realm of the back office to the public's imagination, the public's mind, and it's become something that is becoming more and more ubiquitous as it relates to being used by the public at large.

Jackie Whisman: And still, you write that most companies have not taken advantage of AI, and those who have have benefited significantly. Why do you think this gap exists?

Nitin Mittal: The gap primarily exists, and this is what our research shows, as well as many of the conversations that we have had with the organizations mentioned in the book, is a lot of these efforts started with data science groups. They started with data science groups or centers of excellence that were established, and while it actually helped with the understanding of the topic, it helped with proof of concepts, and it helped with essentially adoption in some, let's say, critical domains of functions, it was not widespread, it was not something that permeated the organization, and it was seen as the realm of the researcher, not the realm of the business manager, and not the realm of the employee workforce at large.

That is now starting to change, and this is where organizations who have gone all in, and the 1% that I mentioned in the book who essentially are, let's say, front and center of how they're adopting AI, how they are leveraging it and harnessing its promise, not only in one particular function or one particular domain, but across the enterprise as a whole, such as DBS Bank in Southeast Asia, or what [inaudible 00:19:52] in China have been now able to progress with, these are the type of organizations that essentially started down a different track, and that track was, how could I essentially improve the top line? How could I be better with respect to how I compete? And how could I enhance the productivity of my employees, and consequently optimize my processes? When they started on that track, the leadership that they brought and how they progressed was the key, what I would call, differentiator of how they've become and why they've become all in.

Rob Atkinson: How much of this then is just vision, leadership, courage to take risks and do something different, or is it more than that, do you think? Are there embedded challenges? I maybe, Jackie, have stolen your next question, but we'll go there.

Jackie Whisman: You did, but that's okay, we were eager to ask it.

Nitin Mittal: Leadership and how a leader, in a sense, uses his or her bully pulpit, if I can use that as a phrase, matters a lot, it matters a lot. If this is organically driven through a data science group, or a center of excellence, or one particular function, it can be effective in certain pockets, but it does not enable the entire enterprise to go all in.

What we found in not only the surveys that I've referred to, but also in all the preparation that we did in the run-up to this book and the interviews that we conducted, leadership has perhaps the greatest impact as it relates to the adoption of AI. It is not the technology, it is actually the leadership that is brought to bear, wherein an executive is able to see the vision, is able to understand the promise of artificial intelligence, and has almost a passion to essentially think of AI more about how they could be creating a difference, and how they could be increasing the means to compete, or how they could be enhancing the experience of their customers, their stakeholders, their employees, and the type of interactions that they actually undertake. If a leader stands up, if a leader drives this as a program, if a leader puts his or her own, let's say, currency towards it, and by currency towards it, I mean political capital, then it goes a long way in the adoption curve of AI within that enterprise.

Jackie Whisman: As we wrap up, any thoughts on the role of government here, what they should do to make AI adoption easier, or maybe as ITIF thinks about it, what they should not do to make AI adoption harder?

Nitin Mittal: Government has a huge role, absolutely, a huge role, because right now, there is a lot of, what I would call, unbridled enthusiasm around this space, and we are certainly seeing that with the number of users who have started using ChatGPT, and that's out there in the popular press. There's a lot of promise, there's a lot of enthusiasm, there's a lot of passion around it, and frankly, even a consultancy like Deloitte have got a lot of inbound queries from our own clients in terms of how they can think about AI in general, and generative AI in particular, and then what type of help we could provide them. So we see that a lot.

But with any technological revolution that takes place, unless there are some guard rails, unless there are parameters, unless there is a regulatory framework, the promise can easily dissipate into something that is not productive for society. And every basically advanced technology, such as AI, while it can help the way that we have been talking off on this particular podcast, can also be used for all kinds of nefarious activities, whether those nefarious activities are from a cyber perspective, whether those activities are in the realm of autonomous weapons, if I can bring that in, or whether that is for, let's say, deepfake, which is also becoming quite prevalent. If not managed, if not governed appropriately by governments around the world, it can have a very detrimental impact on society.

And this is exactly where a company like Deloitte focuses on what we call trustworthy AI. How do you actually trust the applications of AI? And trust has many parameters to it, or many dimensions to it. The dimension of privacy, the dimension of ethics, the dimension of explainability, as well as the dimension of weeding out the conscious or subconscious biases that somehow unfortunately get encoded in the actual algorithms because those are written by human programmers. So there's multiple dimensions to this broader notion of trustworthiness of the artificial intelligence systems that have to essentially be thought through, governed, and regulated.

Rob Atkinson: We're probably slightly more optimistic than you are. I don't really have very much worry about companies that are hiring Deloitte, or any of your competitors really, because everybody has an interest in making sure that these do what they say, that they're not biased, that they're not open to privacy risks, and I think most, if not all, large companies understand how badly they would be harmed if they messed this up. In fact, that may be one reason why they're going maybe a little more slowly, because everybody wants to make sure they're right. I think it's more the companies that don't care or smaller ones, and that's where we've really got to make sure that we limit the harm there.

Nitin, thank you. We have to wrap up. This was really, really interesting, and we're going to post a link to your book, which I'm looking forward to reading. But thank you so much for being here, it was great.

Nitin Mittal: Thank you, and I enjoyed it very much.

Jackie Whisman: And that’s it for this week. If you liked it, please be sure to rate us and subscribe. Feel free to email show ideas or questions to [email protected]. You can find the show notes and sign up for our weekly email newsletter on our website itif.org. And follow us on Twitter, Facebook, and LinkedIn @ITIFdc.

Rob Atkinson: We have more episodes and great guests lined up, and hope you'll continue to tune in.

Jackie Whisman: Talk to you soon.