Good things will come to those companies that don’t wait. On this edition of The McKinsey Podcast, McKinsey AI experts Michael Chui and Alex Singla discuss McKinsey’s new report about the generative AI (gen AI) opportunity with global editorial director Lucia Rahilly. Hear how companies should immediately seize the gen AI opportunity to gain competitive advantage.
But first, a quick US consumer behavior check-in with McKinsey senior partner Kelsey Robinson. This is the first of a series of such check-ins.
This transcript has been edited for clarity and length.
The McKinsey Podcast is hosted by Roberta Fusaro and Lucia Rahilly.
Consumer optimism is up
Roberta Fusaro: Kelsey, how has US consumer behavior changed since this time last year?
Kelsey Robinson: We’re continuing to see what I would call mixed signals from consumers, and it’s driven by an uncertain economy. Inflationary pressure and a tight labor market are causing US consumers to spend differently, and they’re taking actions that are contradictory. There are three main things we’re seeing in our latest research. First, there has been an increase in consumer confidence versus last summer, so now a third of US citizens are feeling optimistic. Last summer, optimism about the country’s economy was at a quarter, and pessimism is waning. That’s good. The second signal is consumers are still worried about rising prices and inflation, pretty consistent with last year. And increasingly, they’re worried about job security. Finally, we’re seeing that while there is this uncertainty, consumers are still opening their wallets. They’re just being creative about it.
Spending and saving
Roberta Fusaro: Let’s talk about that creativity. How is it showing up in terms of how the US consumer is saving?
Kelsey Robinson: The US consumer is choosing to save primarily by trading down. They’re forgoing more expensive brands or stores for cheaper ones. Or they might change the quantity of what they buy to get the cheapest price. That might mean they’re buying less quantity-wise, or in some cases they might be bulk buying to stock up and get a lower price. Overall, and from what we saw in our latest research, 80 percent of consumers are trading down.
There is a category difference here too. We’re seeing slowing growth in categories like home and pet. And the categories that are more resilient are things like beauty and a lot of categories that are linked to out-of-home experiences.
Want to subscribe to The McKinsey Podcast?
Roberta Fusaro: I’m thinking about my own purchasing, Kelsey—just thinking about how much more I am spending on travel and finally getting out of the house. It’s true. How are people splurging, if at all?
Kelsey Robinson: Forty percent of US consumers told us they’ll splurge in the coming year on themselves and their families. Travel is a category they’re definitely saying they’ll splurge on, but they’re also saying they’ll splurge on categories like apparel and footwear.
Roberta Fusaro: What kind of different spending patterns are we seeing generationally, if any?
Kelsey Robinson: We just talked about splurging, and what we see is that intent to splurge is much higher for the younger generation. This is true even if you control for one income level or two.
We have two-thirds of Gen Z saying they intend to splurge and treat themselves. Compare that with 20 percent for boomers. Second, inflation is obviously top of mind, but again, job security is really top of mind, and this is much more meaningfully top of mind for the younger generations. [When] we think about last year, there were lots of headlines on companies tightening their workforces, and we see 74 percent of Gen Z is worried about employment—three-quarters. It’s huge, right? Versus about half of Gen X. And finally, the trade-down behavior is also correlated with generation. So there’s a dichotomy. While Gen Z tells us they’re more likely to splurge, they’re also telling us they have to actually trade down and manage costs in other areas.
Companies should be flexible
Roberta Fusaro: If folks are changing quantity or trading down to cheaper brands, what are the implications for companies?
Kelsey Robinson: There are three main implications. There’s a real need in this moment to find the way to deliver value and meet the consumer where they are on the trade-down side and communicate how you’re delivering value, because every consumer is thinking about it. At the same time, again, there’s that spend dichotomy. There’s a need to figure out in what moments, for what products, is the consumer willing to splurge. Are they willing to treat themselves? Can you be part of that joy?
Number two, the idea of really personalizing how you’re speaking to consumers in this moment is critical. You and I probably have different purchase behaviors right now, so how do companies learn as much as they can to make sure they’re meeting the consumer where they are? Over 70 percent of consumers expect companies to personalize things. They get frustrated when it doesn’t happen.
And finally, one learning we haven’t touched on yet is there was a big change—since COVID-19 and even since last year—to omnichannel shopping. Over 80 percent of consumers are researching and browsing across multiple channels. That’s six percentage points higher than last year—whether they browse on the web or in the store and then convert in one of those channels for a single trip.
What’s different now is it’s not just the younger generation; it’s also Gen X and baby boomers who are doing it. So continuing to figure out how to serve and create a seamless experience in an “omniway” is critical.
Roberta Fusaro: Kelsey, thanks so much for joining us.
Kelsey Robinson: Thanks for having me.
How generative AI can benefit business
Lucia Rahilly: Generative AI has exploded into the popular imagination over the past ten months or so, catalyzed of course by the launch of ChatGPT last November. Many folks are ricocheting between delight and awe and probably in some cases fear at applications like ChatGPT, in part because they’re so simple to use and, frankly, uncanny. Alex, what did this new research tell us about the business areas that are likeliest to benefit the most from gen AI?
Alex Singla: As we think about this research and the impact it can have across businesses, it’s important to recognize that a lot of this has been around for some time. But as you mentioned, it’s getting into the mainstream. As we think about where different opportunities can be created, compared to previous automation research, gen AI has real influence on white-collar work.
So activities in the banking industry, in particular, and the healthcare industry have huge opportunities that one might not have seen in prior technologies and prior automation, given the white-collar knowledge worker.
Lucia Rahilly: What is it about these specific industries like banking and healthcare that makes them so conducive to benefiting from gen AI?
Michael Chui: For instance, marketing and sales has an ability to either create personalized marketing or, in some cases, even create brand marketing to create an advertisement for you, a script, the music, even video is coming at some point.
These technologies are also good for customer service or customer operations—for example, chatbots to the extent to which they can actually answer questions.
Another area, which we spend a lot of time on, is the potential of these technologies to improve the ability to write software, because to a certain extent, computer languages are just languages. These are, you might have heard the term, large language models. You can ask it in English or whatever language to, “Please write me a piece of code that does XYZ.”
We have a lot of programmers in banks. We have a lot of programmers in retail companies. So that’s a large potential for value as well. The most underrecognized case is the ability of generative AI technologies to do generative design in research and development. For example, you can create an electrical circuit or design a car.
There’s no business nor function where these technologies don’t have some potential.
Alex Singla: On the R&D side, in the life sciences industry, I’ve heard executives discuss chemicals and the different configurations you put chemicals in to create new drugs and new products that would take years to experiment with.
Can they leverage generative AI to put different chemicals together to accelerate R&D in the pharmaceutical space? That hasn’t been done or proven yet in terms of making the perfect drug or product that much faster. But that’s the kind of thinking that’s happening.
Gen AI augments imagination
Lucia Rahilly: Do you think it’s possible that generative AI will expand the frontiers of creativity and innovation in new ways?
Michael Chui: Yes, if you can use these systems to accelerate productivity and create more examples. What we know from our product development research is that, oftentimes, the problem is you close the funnel too quickly. You don’t continue to look at multiple alternatives.
Clients are saying, “We can create more alternatives using generative design.” And if you can hold open more potential, then maybe you can end up with better products at the end of the day, create whole new categories, create new drugs, create systems that are more sustainable and produce less carbon emissions. All the kinds of things we want to do with R&D, we can create a superpower by using these technologies. It can be great not only for companies but for humankind.
Alex Singla: One other thing on the creative side is, yes, generative AI can be creative, but it also increases the creativity of the user.
Let me tell you what I mean by that. We had a roundtable of chief marketing officers in Europe. We gave them a one-minute exercise that walked them through making their own creative for a product within their business leveraging generative AI.
And you just saw their eyes open up. They were familiar with gen AI’s potential, but it was giving them ideas they might not have had before. Even if the answer wasn’t perfect at the end, it was expanding their thinking on what was possible. So generative AI can accelerate your own individual creativity.
How gen AI can generate productivity
Lucia Rahilly: Let’s turn now to attach some numbers to all this and talk a bit about what gen AI might really be worth to the businesses that harness it successfully. What kind of value might gen AI add to the global economy?
Michael Chui: When we look across all the different use cases in corporations and other organizations, we’re talking $2 trillion to $4 trillion annually of potential value that can be unlocked by using these technologies.
Every knowledge worker has the potential to use these technologies to increase their productivity. If I can have something write the first draft of a document for me or an email, that accelerates my personal productivity.
And if you look at all those micro cases, the actual potential might be double of what I just described at the corporate level, because from a macroeconomic standpoint, we’re all aging. In fact, we won’t have as many people. If we’re going to have the next generation do better than our generation, we really need to increase productivity. That’s one of the potentials we have in front of us.
Lucia Rahilly: So I’m clear, are we talking about gen AI specifically on top of the productivity value that we expect regular artificial intelligence to bring, or both together?
Michael Chui: The $2 trillion to $4 trillion is gen AI only. But when we look at the overall economy, there are lots of other types of technologies, including physical robotics, and other types of AI, which also can generate value. And we shouldn’t forget about those, particularly in the corporate context.
Alex Singla: One other thing that distinguishes generative AI from other technologies is that it’s so simple to use. One of the reasons things took off is because the user interface was just so darn easy.
Whether you’re eight years old or 80 years old, anyone can use ChatGPT.
The willingness to adopt is much greater than before. One of the challenges historically with AI is people described it as, “Oh, it’s a little bit of a black box. I’m not exactly sure what those people in the data science environment really created there.”
Michael Chui: To be clear, eight-year-olds often could be faster adopters of many technologies.
Alex Singla: Fair enough.
Lucia Rahilly: I have one at home. I can attest to that. Faster than I, anyway.
C-suite in the gen AI game
Michael Chui: But it’s supported in recent survey research that we’ve done. We’ve been serving thousands of business users around the world on the use of AI.
What’s interesting is this year we asked about their personal use of these technologies.
Roughly the same percentage of people in the C-suite as in middle management, as in entry-level positions, had already started using these technologies on a regular basis, both at work and in their personal lives.
Oftentimes, some technologies are mostly used in middle management or on the front line. And executives have other people do that stuff for them. But here they really have personal experience, and that’s quite powerful.
Alex Singla: I had a Fortune 500 CEO client who blew me away because he said he’d been taking an AI class for the last six weeks on a Saturday, and it was an eight-week class. He himself was spending time learning how to code, what it means, and what it takes.
This is just an example of how executives are thinking about their own learning journey along this dimension of generative AI, which I wouldn’t have heard a year ago from clients.
How gen AI can help reach consumers
Lucia Rahilly: In the research, you suggest that in retail and consumer goods gen AI could increase profits by as much as $400 billion to $600 billion. What opportunity do you see for specifically those industries?
Michael Chui: If you think about sales and marketing, retailers, or a consumer products company, you want to get to consumers.
There are a number of ways to do that. One is if you do brand marketing. Alex’s story about the CMOs is very instructive here. If the marketing department can create more alternatives, more first drafts of brand marketing scripts, video or images, music, that accelerates the productivity. Can they get better, more powerful messages out faster?
But also, you want to go straight to consumers on a one-to-one segment, “one type of marketing” basis. You want the ability for these technologies to create more compelling, individualized marketing that creates value.
Similarly, in customer operations, you often have customers calling in. And you have people helping in your contact centers. You can automate some of that to make sure there’s always someone available to talk at 3 a.m. when you have a question about defrosting the turkey.
You can actually use these systems to augment human contact center representatives as virtual experts. That’s one of the meta use cases that we know these systems are good at. If you can take a corpus of your corporate data and hook it up to a large language model, you can query it.
Alex Singla: Another area in the retail space that I’ve been having lots of discussions around is store operations. Often in store operations or in fast-food-chain operations, the manager who’s running that operation could provide the best-practice insights for how to use generative AI tools around key store operations, whether that be inventory, sales, or footprint.
Having those tools at your fingertips to make better business decisions every day is an enormous unlock for retail operations, especially if you’re a large retailer with a thousand stores across the country, where the variability of that store or operations manager is quite large.
Michael Chui: There’s an interesting piece Joanna Stern from the Wall Street Journal did about drive-through lanes in quick-service restaurants, where using this generative AI technology allows a customer to have a conversation with a chatbot.
Lucia Rahilly: Wow, OK.
Michael Chui: Exactly. You can ask, “How many calories are in this? I’d like to modify my order in the following way.” And she was testing it. She had recordings of screaming children and dogs barking. She laid on the horn during part of it. And, by the way, if it has a problem, then it turns you over to a human too. You can make the process resilient by adding different layers.
Restaurant operators don’t have enough crew to serve the drive-through lane. So the use of the AI bot didn’t reduce any labor. It just allowed them to provide the service, which they otherwise couldn’t do.
The elephant in the room
Lucia Rahilly: To that point, which jobs do we expect to be affected the most? What does gen AI mean for the future of work and for workers?
Michael Chui: In our research, we use a taxonomy that details roughly 20 to 30 work activities per occupation. Typically, what these technologies do is automate parts of people’s jobs. You could describe that as creating superpowers.
Because if the machine does “X” and saves me an hour, now I can use that hour more productively. So I think that’s the number one thing to think about. Number two is, in many cases, and this came out in other research that we did on the future of work in America, for instance, despite some parts of people’s jobs being automated, the demand continues to increase.
Let’s take the example of software. I can increase the productivity of software engineers, but we still need more software. We’re doing this at our own firm. So we’re going to accelerate the productivity of the people we have. We’re not going to have fewer people. We’re going to keep adding people and having them be more productive, create higher-quality software, more software. But there will be shifts. Indeed, there will be shifts over time.
Any individual management team will have to make a decision about what you’re going to do with the time that gets freed up. Some people have talked about a four-day work week. But in fact, between now and 2030 in the US, for instance, about a third of the activities we’re paying people to do today potentially could be automated, not just using generative AI, but all kinds of technologies. It’s a fast pace but not that dissimilar to what we’ve seen historically.
Lucia Rahilly: What should knowledge workers be doing now, if anything, to prepare for this new era of gen AI? What kinds of new skills or muscles might we have to develop to make this shift?
Alex Singla: If you’re a knowledge worker, start playing with the tools and understand what they can do. Because what it allows you to do is to take away some of the mundane tasks you don’t want to do anyway. It frees you up to spend more time problem solving, cracking hard problems, spending time with your people, coaching your folks, and really thinking.
I think there’s a shift in terms of how people will spend their time over time on the knowledge worker side. One thing I always encourage people to do, just like we did with the chief marketing officers, is play with it so you can understand how you can leverage it to actually be more impactful, more effective in what you do.
The time is now
Lucia Rahilly: You mentioned the marketing executive example. What about business leaders generally? What should business leaders be doing to make this pivot?
Alex Singla: Continue to play with the tools; understand their uses. The reality is, it will require people to retrain, learn new skills, redeploy themselves in ways in which they can find and create value in different circumstances.
Michael Chui: Our colleagues wrote this book Rewired recently addressing all of those aspects of how you transform a company using technology. Gen AI is another tool that allows you to do that.
Alex Singla: A question that comes up for a lot of the organizations I’m working with is, “How do I redistribute that to do the things on innovation, on R&D, on new markets that they didn’t have capacity to tackle?” It opens up the sphere to tackle things you didn’t have time to do before.
Lucia Rahilly: How will companies sustain the competitive advantage they get from gen AI? In other words, will gen AI simply become table stakes? Or will some leaders use gen AI differently from others in order to sustain that edge they’re gaining?
Michael Chui: Sadly, in some ways it is a run-faster business—the “if you get there sooner, you’ll move up the learning curve” business. Everyone is running. But generative AI isn’t going to give you a competitive advantage forevermore.
What we’ve found is these technologies tend to increase the spread between the leaders and the laggards. If you learn faster, you can achieve a longer-lasting competitive advantage. But that’s not because you do something once and then you rest on your laurels. You have to keep learning, pushing, adding to the capabilities and effectiveness you already have.
Alex Singla: There are certain areas that, to your point, absolutely become table stakes over time. Let me give you a simple example.
I’m in human resources, and I’d like to have generative AI write me job descriptions. Will that become competitively taken away across every company out there? Of course, it will. That will just become core to any core HR technology platform that people end up buying over time.
However, there’ll be other solutions and opportunities with generative AI, in which I’m leveraging my internal proprietary data and insights into the model that will give me strategic advantage over time. The better I am at leveraging my own data and insights into the model, the more it can be a competitive advantage.
Speed is often a strategy in and of itself. So those who run faster will win over time. At the same time, those who learn faster will also win, because they will be able to deploy the solutions and get greater adoption. Getting people to adopt the change management component is key. It has less to do with technology and more to do with people. It will be another major factor in how you create strategic distance to win over time.
Adoption is global, value is local
Lucia Rahilly: Anything in the research on geographic variation? How much traction do we expect gen AI to get globally?
Michael Chui: What we’re seeing is that it’s being adopted all around the world. Partly because of the ease of deployment and the fact that these technologies often speak many different languages, computer languages included.
Alex Singla: One thing I find fascinating on this topic of geography is how different use cases are being adopted or should be adopted based on region.
For example, let’s take a generative AI tool that would give recommendations to a patient on what they could do if they had certain symptoms. Usually in the US or Western market, you’d need a doctor in the loop between that machine and the patient, because you’d want to make sure that from a liability perspective you had confidence in the patient recommendation. But if you look at a small village in rural India or Africa where the person doesn’t have access to a hospital or doctor, then all of a sudden the US solution might not work due to the speed of deployment.
The solution is going from zero to 100 with less viability, but that’s a better alternative for the patient in, for example, India than you might see in America.
I find it fascinating to think about how generative AI can add value and how it might differ depending on where you are around the world.
Michael Chui: I want to emphasize a point Alex was making, which is as we’ve been talking with clients, we found it’s really important to work on real problems, not just the little toy problems, but to grapple with the hard problems that can drive value. Because, as Alex was saying, you learn things and, to the extent to which you have limited resources, limited talent, use it on problems that really matter.
Lucia Rahilly: How quickly do we expect these changes to begin to take hold and really start to have impact?
Alex Singla: It kind of depends. The simple use cases will go fast.
More complicated use cases in which you’re going to leverage generative AI to, for example, reimagine or transform a customer journey or a domain of your business will take more time to get implemented, because it’s going to force you to tackle some of the really hard and realistic issues and questions, whether that be risk on data privacy or IP [intellectual property] protection. Those are real questions. Questions around the implications of “What am I going to do if it has people implications? Am I going to capture the value or am I not?” need to be considered.
You can outline your guidelines and protocols on those, and you should, but that’s a little bit in theory. Until you start to apply them in a real-world use case or domain or journey in which you’re forced to make these decisions around risk management, legal issues, people issues, technology issues, that’s really where the rubber will hit the road.
Lucia Rahilly: So how much time do you think folks have?
Alex Singla: I think six months ago, people were kind of waiting and seeing. But the jury has made a decision, and I would encourage companies to get moving, get learning, get having impact, and build excitement. That in itself will create a flywheel of speed and impact.
