How can asset managers navigate today’s rapidly changing technological landscape and gain market share despite economic uncertainty? The generative artificial intelligence (AI) revolution is well underway and is already transforming the way asset and wealth managers operate and are delivering significant efficiencies.

In this special edition of our Oliver Wyman podcast series, Joshua Zwick and Kamil Kaczmarski, partners in our Insurance and Asset Management Practice explore The Generative AI Tipping Point, providing analysis and interviews with senior leaders from our global wealth and asset management report. Joshua and Kamil discuss the asset management industry’s outlook and specific actions to capitalize on opportunities. They share the revolutionary power of generative AI and how it can be implemented to drive better investment decisions, engage clients more effectively, and enhance productivity across the value chain.

In this podcast, Joshua and Kamil discuss:

• The state of the global asset management industry at this pivotal moment.
• The inflection point we call ‘the big sort,’ and what it will take for organizations to land on the right side of this reshuffling in the industry.
• Specific actions to reimagine their operating models and build greater resiliency in their businesses.
• The levers management can pull to drive profitable growth and operational resiliency.
• What makes this moment a tipping point for generative AI and how generative AI is already impacting business and society.
• How and where asset management leaders might deploy generative AI to reduce costs and gain competitive advantage.
• Unlocking the power of generative AI for building talent, processes, and business models, and how teams can begin to access its impact.

Joshua Zwick: Hello, and welcome to The Generative AI Tipping Point, a special edition of our Oliver Wyman podcast series. Today, we're going to dive into the global asset management outlook: What's trending in the industry and how managers can win market share and drive growth and build operational resiliency, and then we’re going to pivot to the hot topic of the year—generative AI—and what it means for the asset management industry.

I’m Joshua Zwick, a partner at Oliver Wyman who focuses on the asset management industry. I'm joined today by my colleague, Kamil Kaczmarski, a partner from across the globe based in our Frankfurt office.

Kamil, are ready to talk about the report, The Generative AI Tipping Point, and all the interesting findings that we have written about?

Kamil Kaczmarski: Absolutely. 2022 represented a wake-up call for asset managers. After years of buyout that allowed us to hide underlying fragilities, 2022 highlighted how sensitive asset managers are to market downturns, especially on the traditional active side.

The beginning of 2023 was a bit of a recovery and a rebound in the market, so it helped to ease the sting a bit, but the future environment still looks pretty much challenging. We have uncomfortable high inflation and still a lot of geopolitical tension. So, all of this is actually fueling instability and a little bit of uncertainty about the outlook for growth and especially wealth creation.

I think right now we are in an inflection point, an inflection point that we call the ‘big sort’ for asset managers. There will be a reshuffling of priorities, but also a shuffling of competitive forces. These conditions will exaggerate the differences for managers that can navigate these conditions and be on the right side of the ‘big sort’ and those who can't. And actually, those who can’t, will find themselves in an unhappy position.

So, the question for asset managers right now is, ‘what do I have to do to be on the right side of the big sort?’ What are the winning strategies from those changes in the market? This is what we have talked about in our global wealth and asset management report, and we will discuss today in this episode.

Josh, if you look back at the findings of this report, what were the biggest surprises for you?

Joshua: I think the first thing that stands out is how generative AI is no longer on the whiteboard. It's not just a bunch of people standing around talking about ideas. It's actually being used in business. It's coming out of the laboratory into the production lines and that's creating real impact. We'll talk about this today, it's really much more in the front office and in client-facing areas — where it's having the biggest impact. And we're seeing huge impacts in the order of 20% or 30% in terms of efficiency gains and in particular other areas where it's even significantly higher. So that's going to be a big area and something that is moving really, really, fast.

The second thing that was surprising is the huge opportunity in traditional active. We all know that the secular dynamics of the flows going from traditional active to passive that's still continuing at a pace. That's not slowing down. The analysis we did, which we updated from a couple years ago, shows the trend is still very strong if not stronger, and that the actual flow between active funds, from active-to-active funds, is more than three times higher than the flow from active to passive funds.

So, what does that mean? What it means is that there's actually a tremendous amount of money up for grabs. Money in motion if you will. So, while the pie might be shrinking, if you can get a larger slice of that pie, it's actually a huge opportunity, particularly when we think about kind of the revenue weighted perspective because passive, as we all know, there's not that much money to be made unless there's massive scale there.

Kamil: I love this analysis because I think for a long time there was the ongoing debate about whether active is dead. I think if you look at the analysis and the numbers, it’s positive news and there is still enough revenue to fight for. But Josh, if you think about the kind of levers those managers can pull, what will come to your mind?

Joshua: Yeah, of course, if you have a great performing fund or fund lineup, and you have in-demand strategies, that's always going to be a great thing, right? But as we've seen time and time again on average, we don't see the active, at least not in the traditionally active world, generating significant performance above benchmark.

Obviously, everyone wants to have better performance. But there are other levers that that you can pull and three came out in the research. The first is around product innovation. When you actually look at the data, what we found is that newer funds in terms of the flows that they attract are significantly higher and do significantly better than older funds. I know there are some structural reasons for that. There are marketing dollars put against newer funds. Those newer funds ideally would address a particular market demand at that time. Structurally, they don't have long periods of underperformance where people will be redeeming. But with all that said, what we're seeing is a real advantage to those organizations that can actually put good product out there that meets the demands of the time. And so, I think that's one.

The second was around distribution in service. And we talked about distribution alpha, service alpha—this is real thing. So, from the data we looked at a wide variety of strategies and grouped them by performance and we actually zeroed in on those that did the worst, right? The bad performing strategies across a large number of managers. And then we classified those into, well, if you had bad performing strategies, what's the difference between those managers that did well in terms of retaining AUM or actually driving additional flow versus those that didn’t? We saw a huge skew between those that had really good flows despite the bad performance relative to those that didn't. And you'll see strong distribution networks obviously make a difference. The service levels that are being able to provide it, whether it be solutions and advisory capabilities that are kind of overlaid in terms of the product, that creates a degree of indispensability for these managers to their clients and that shows up in the data.

The last bit of analysis that I thought was interesting as well — you can call these all different forms of alpha, there's performance alpha, product innovation alpha, there's distribution in service alpha. And the last bit of additional non-performance alpha that we saw was around fee structures. And this was also interesting. The funds that had lower fees garnered a significantly greater share of flow. So, what this is suggesting is that investors like the promise or prefer the promise of lower fees relative to the uncertain promise of outperformance at some future point in time. So, being really strategic about how you think about fees makes a big difference because even if you have bad performers, those that had lower fees typically did much better across the dimensions around collecting AUM and retaining assets.

Kamil: Super interesting.

Joshua: It is. So yes, there's probably one or two other things I thought stood out, Kamil. One was actually around pricing, particularly institutional pricing. And we think that managers are leaving a lot of revenue on the table. And so again, we try to let the numbers do the talking here. And we know that published fee rates are significantly higher than what ultimately organizations end up paying. What we found here is that we have some information that shows there are massive skews in terms of the price paid for the same mandate, for the same size of that mandate, across the client base. So, you look at the skew and see some have paid significantly more or significantly less for the same product at the same size level.

If you narrow that band and be more selective in terms of how you manage the discount and pricing process, we see a tremendous amount of money that's being left on the table. But we do think that this is a lever that organizations or asset managers can pull. Being much more disciplined how they think about pricing strategy around client value. How they utilize data to really understand what the long-term client value is, right? How the incentives are aligned and avoiding a situation where salespeople will give away anything at the point of sale to get that gross revenue in the door and think much more strategically about how to price the product, with the broader context of the relationship obviously and the kind of long-term value of that client. And that’s not an easy thing, right? It’s not an easy thing to do. But what we’ve seen for those that can do it quite well, there’s a significant advantage in revenue uplift that you can get from kind of more effective and a negotiation tactics.

Kamil: Interesting. What will be your final surprise?

Joshua: The final surprise is, as you know, we've talked for a long time about the importance of and the growth engines of private markets and an alternative to some extent, and we still see this. That's going to be a key driver of growth going forward.

And we've seen a lot of activity from traditional managers looking to get into this game in many cases inorganically. And I think while that can provide a lot of benefits and tap into a growing market, we have also seen huge operational challenges associated with that. So, if you actually look at the benefits of bolting on a third-party, an inorganic kind of private market strategies or teams or business into the kind of the core franchise, what we see is there can be a lot of friction, whether that is kind of operational-oriented friction, whether it is going to challenge in terms of kind of unifying platforms either from a technology perspective or distribution perspective. That can actually weigh on the overall kind of operational efficiency of the business. And so, the point being, it's not one of those things where you can just kind of bolt it on and everything's all good. It takes a lot of work to successfully integrate and actually achieve the benefits that are the basis for those types of strategic acquisitions.

So, I’m interested, Kamil, in terms of some of the things that that you might have taken from all the work we did together.

Kamil: I think I’ll highlight two topics. Connected to what you just talked about with traditional players expanding to alternatives, what we said at the beginning about the ‘big sort,’ I would expect, some sorting—rebalancing—within private markets. So, it’s definitely one of the key growth zones.

But if you double click private markets and then look at what's happening underneath, there’s some very interesting dynamics. If you think about real estate, this is an asset class where there's still a bit of concern in the market. With this asset class, it's still a rebalancing of the attractiveness, relative attractiveness of real estate.

Private debt is definitely right now people's favorite, I remember one of the executive interviews when a person said it's like Christmas for us. It was a private debt player. Let's see how long, but it is definitely an area where we see not only in corporate private, but also in real estate where the yields are so attractive versus equity, real estate equity.

There’s a lot happening where those who didn't have experience yet are entering this space and others are simply realizing, especially right now in an inflationary environment, it is actually a very attractive asset class.

And then, finally, two of my favorites: infrastructure and natural capital. I think both of them are strongly benefiting from the entire trend we have right now in responsible investing and especially as an asset owner, are finding suitable investments that will help achieve ambitious targets. Many are pushing on the boundaries of net zero. You will need a lot of support from investing into infrastructure and especially natural capital such as forestry, to reach those ambitious targets.

So, as depicted in one of our charts we show how the revenue margin changed over time for the industry and the cost margin. There was a very clear drop on the revenue side. But the cost side remained stubborn. There was little change on the cost side. And this is the over all problem we as an industry are facing — that we are not yet entirely prepared for a potential market downturn.

I think some of the players were stuck in the middle, especially where they probably lack the scale. They will be hit hard if we enter a market downturn that will last more than simply just one quarter.

So, what we did in this report and what I really like is our menu of different levers that asset managers can pull. It’s a healthy check to go through them and understand where you are. And if you just look at them, there are some that many managers have addressed or are looking into. These are the typical cost ranges I would say 5, 10, 15, up to 20% of your cost base. But if you really want to think hard about the potential of really reducing your cost base beyond the 20%, this is where you have to make some really tough choices. Tough choices about your structural cost base. And tough choices about questions, such as your location footprint. Do I have to be present in all those locations and the satellite offices? Tough questions about your investment landscape or the capability landscape. Is it credible for me to, for instance, manage Japanese Yen out of Europe?

I think the industry is right now at this inflection point, but those hard choices are on the table. There's an scenario out there of a longer downturn and we will have to operate in an environment where the cost has to be adjusted.

Joshua: Yeah, and these are perhaps somewhat provocative statements and the possibility to get 5% to 15% or around 20%, but there's a lot more on the table if managers are really willing to make those hard decisions.

And so, one of those things that we’ve talked about, obviously a lot in the report and I'm talking about now, is what's the role of generative AI in this? I thought, Kamil, I'd walk the audience here through some of our perspectives on Gen AI that we covered in the report, and a keyinitiative that’s possible is changing the operating model to drive significant efficiencies. But that's just part of the story, right?

So perhaps, if you don't mind, we’ll shift over to the Gen AI side of the discussion.

Kamil: Let's do that.

Joshua: Alright. So, one of the things we've coined is the term the ‘big sort,’ and hopefully that is evocative of the way that we see the industry shaping out. The other thing is where we are as an industry and this concept around the generative AI tipping point, and why this time is different.

And we see the generative AI tipping point being driven by three things. There's three aspects. The first is the higher degree of accuracy. And you know, if you looked at how GPT-1 or GPT-2just responding to a simple email, what resulted was pretty bad. It's clearly not coming from a human being. So, if you are using that as a chatbot it is not good. If you look to take an exam, like say the Bar Exam, I believe GPT-3 passed it by 10%, but GPT-4 passed it by 90%.

Greg Jensen, co-Chief Investment Officer of Bridgewater Associates was on a podcast recently and he commented that GPT-3.5 was able to pass the investment associate exam given to all investment professionals at a 20% level. With GPT-4, guess what? 80%. Having an infinitely scalable analyst similar to those from the top universities at your fingertips is pretty remarkable, right? So that accuracy is just crazy.

Second is the broad range of the application, the generative AI applications, nowadays. And it’s not just about generating text and creating new ideas. It's also obviously creating new images or new music, but I think what's remarkable is the generality of it. It can be applied to revolutionize science, right, and being able to review journals on protein folding and come up with novel new molecules, right? Or review massive amounts of investment research. Or it can just be used to create a really clever holiday card, right? It is amazing how broadly applicable these types of applications are.

I think the other thing, and this is probably the most transparent or salient for most people, is the accessibility and user friendliness of interfaces. Anybody with a high-speed internet connection and maybe a couple bucks in their pocket to go to GPT-4, you have complete access to absolutely groundbreaking technology, right?

So, you combine those things — higher degree of accuracy, the broad range of kind of applicable use cases and the accessibility, the user friendliness — it really is a game changer.

Kamil: And within the asset management industry, if you think about the value chain, what do you think would be the largest opportunities?

Joshua: Yeah, so based on our research and talking to a large number of asset managers in this space, I think the first point, I alluded to this earlier on, is that the revolution is well underway. These are not use cases where they're not just on some kind of paper or on a whiteboard. Organizations are actually doing. And what we see organizations actually doing is focusing much more on the frontend where there is client interaction, where there is the investment process to some extent — I'll clarify that in a second — and areas where there are huge advantages that the nature of Gen AI can actually create. I think this is important point.

So, before I get into the specifics, I just want to make a one comment which is that Gen AI is not perfect for everything. It’s not the panacea, right? There are certain types of things it does really well, certain types of things where other types of technologies or process redesign or other types of machine learning models are much better for. But when you have a situation where there's some degree of routinization, something that requires the creation of new content and to some extent customized content, and to the extent that there's a large amount of information to synthesize, if you have those conditions, Gen AI is a really good tool to utilize in those situations to generate massive efficiency gains.

And so where we see those types of situations being most prevalent, it's often in the front office, client facing, to some extent the investment research areas. And so, think about writing, transcribing, and synthesizing calls. There's really no need for human to do all of that, right? Think about spending time drafting insights about clients or talking points to support targeted sales. There's really no reason for human beings to do all of that? Think about drafting RFP responses or customizing marketing materials or thinking about synthesizing research reports, right? Or reviewing and summarizing mountains of due diligence documents as part of a deal. Yes, you still need humans in a weigh in, and we'll get to that, but a tremendous amount of progress or efficiency gains can be made by not starting with a blank sheet of paper. Imagine we've all been in a situation where we're looking at a sheet of paper or a mountain of documents to synthesize and go, where do I even start? Gen AI can actually get you way down the field to use a football analogy, sorry, Kamil, and to the 50-yard line and well into the end zone in some of these cases, right? And so that sort of uplift, we think is tremendous. Now, we don't see it being used anytime soon for thing such as picking stocks.

Kamil: That will be my next question. It will be my next question because if you would tell me right now that you've found the tool that will tell you what stocks to pick or how to push in the market, well, Josh, believe me, I think we would be in the Bahamas sipping pina coladas, but tell me more.

Joshua: I would agree with you, but Gen AI is say, 80% right in some cases, much less in other cases, right? We’ll talk about the limitations shortly. And what often generates alpha is the last 1%. And so, the generalized answers that it provides are obviously not overly useful for that. But for developing new ideas, new investment concepts, identifying potential risks, things you may have otherwise missed just because of the capacity constraint. You can read and research so much.

Those are the types of things that can be supportive of an overall investment process. And so, it's not going to be used,to trade, build entire trading systems off of it. But it can be really valuable as inputs into the overall process. And that's really kind of what we're seeing and where you think the greatest benefits are likely to be.

Kamil: And if you think about the kind of elements or prerequisites to be successful, is there anything to highlight? Is it for everybody or are there any elements you simply need to have in place to really harvest this potential?

Joshua: Yeah, I know that's a good point. And there's no doubt there's a degree of democratization that this technology provides. I can actually now write code. As long as I know that the questions I want to ask, and I have some understanding of what I'm trying to achieve I actually don't need to be that skilled in C++ or Python to actually write the code.

And so, there's an element of this that actually democratizes things. Any time you democratize something in an industry that is largely commoditized and is looking for that small edge to generate alpha, something that's democratized is actually not all that useful. And so, while it can provide a lot of these efficiency gains that we're talking about, it really just kind of raises the bar. Everyone’s going to have to do that. And I mentioned the speed, this is moving fast.

And there’s a couple of success factors that we call out the paper and I'll just kind of highlight here. One is the proprietary data. We're not going to have institutions creating their own large language models (LLMs) and training these models. It's just way too expensive and way too complicated. But to the extent that they can be fine-tuned to capture the tonality of a particular way in which an organization communicates to capture all the insights from all the emails, the research in the reports, the market commentaries, and the client dialogues, all of those things, those sort of insights, those nuggets can be harvested and combined with the capabilities that these large language models have, by being trained on basically all of humankind’s knowledge.

It's actually allows you to do much more interesting things that are really specific and harness specific kinds of capabilities, insights, research, , if you will, of these organizations. So, in a way it becomes less of a data science question and in many cases more of a data engineering question: How do I get all this data together to basically tune these models and incorporate these sorts of insights into something that is really useful and arguably differentiated relative to all others?

Another key differentiator, Kamil, is probably more what we like to call a systems-based approach. And I think while there's a tendency to write down a lot of use cases and we've been with some clients who have probably hundreds of potential use cases. How do you think about them more holistically? What's a strategy that can be applied to really focus on the on the return on investment (ROI), because that's another element of this, which a lot of people are very excited about the prospects. And hopefully, from my voice you can detect how excited I am! Inside Oliver Wyman we've created some applications that I'm currently using that are making my life significantly easier, including some of the more administrative tasks.

So, I'm really excited about Gen AI, but we need to think strategically about what the return on investment (ROI) actually is because there is a cost. And in many case the cost isn't all that high, but it is high in terms of people's time. And so being clear in terms of what's the overall strategy, how do we think about this more systematically across the organization? How do we allocate resource in the best possible way? That can make a huge difference. And I think that can create some sort of systemic or systematic advantage over time.

The last thing that is potentially a differentiator in terms of how you can deploy these capabilities is reimagining the operating model. And one of the big changes you often hear in these types of conversations around the applications of Gen AI is that it changes human beings from creators to validators, or to reviewers. And it's not to say — we still need to create good content, but ultimately the models need to be trained and as an industry, we're facing some of these challenges at the moment. But being able to recognize the changes that these sort of models and technologies create. If we are doing much more reviewing, does that mean we need to have a center of excellence, in terms of reviewing things that go out to clients. How do we ensure that we keep humans and the right humans are in the loop because we're probably not at a situation yet where you just literally are letting one of these AIs go into the wild and actually provide advice, for example, or detailed information to end clients.

Organizations need to understand how the roles will change and how to centralize things. Do we need to decentralize some things? Do we need to change our processes? Those can create sustainable competitive advantages, but it really requires understanding how the technology impacts the way the organization operates.

And the last I'll point, and there are others, but this is around the data governance and controls. Organizations have information that's sensitive. Secrets for their own kind of trading strategies. Other organizations obviously will have private information about individuals, whether the employees or clients. And letting that of seep out is a real problem. And this creates a big perimeter or surface area for malicious actors. And so, I think those organizations that take a really sound approach to managing the data governance and the data security are going to be essential to making responsible use of these technologies.

Kamil: Super interesting. So, Josh tell me, what do you think will keep our clients busy for the next couple of months when it comes to Gen AI.

Joshua: One is around this systems-based approach to thinking about their generative AI roadmaps. And so, we see a lot of organizations putting a lot of ideas out there. But how do you actually prioritize them? What sort of frameworks do you want to use? What are the criteria for where this is a good idea versus where it's not? Where is the ROI actually worth it? And so, a lot of organizations are struggling quite honestly with the prioritization and how to create more of a systems-based approach to deploying this type of technology at scale throughout the organization.

The other area is around execution implementation. Like I said, this out of the lab and it's into the system. It's into production right now. And so, we spent a lot of time developing tools internally to support our business and our ability to support our clients, but helping our clients actually develop their own tools, host their own applications, and actually train in a way that allows the organization to generate proprietary insights and something that’s differentiated that only that organization can do relative to all others.

So really that execution and implementation, that's where organizations are moving to and that's the types of calls that we're receiving.

Kamil: Excellent. Josh, we are about at the half hour. Any final words of wisdom as a closing remark?

Joshua: Well, I've been doing a lot of talking, Kamil. I'm tired of listening to my voice, so maybe I will put you on the spot. Anything else that you wanted perhaps to comment that we’ve missed or any closing thoughts for our audience who made it this far with us.

Kamil: I think I would encourage everybody to try and pilot the potential of generative AI in their organization. You can start small, pick an area where you understand there is a lot of manual intervention. There's a lot of creative work that's being done. And as you know, my history before I joined Oliver Wyman, I started my career in product development. And I have to be honest that I caught up with some colleagues from a different company and they told me how they do the process today and it is still the same. I think Gen AI gives another opportunity to really rethink how you get things done, especially the area, as an example, product development, right? When it comes to writing a narrative, a sales narrative. Or even the repetitive process of launching a product where very often you have to go through internal committees, and later you have to go to the regulators, and every time you just need to compile various documents or text Gen AI can help you create efficiencies. So, I would encourage organizations to build use cases and try it out. Because the potential, once you see it, is there already. I think it's eye opening. To see how much potential that we have and how much it will keep us busy for the next couple of years.

Joshua: Well, I think that is exactly right. I actually love playing with AI. As long as the organization keeps in mind the things we’ve discussed and considers how to make it successful at scale. That’s where I think those are going to be on the right side of the ‘big sort’ versus those on the wrong side.

So, with that, I hope everyone enjoyed our conversation today. I've been involved with our Global Wealth And the Asset Management report for many years now, I think this really is the best one we've done, so I encourage all our listeners to read through it and hopefully they will gain unique insights out of it.

We're also doing another podcast on the wealth management side of report. So, my colleagues will be leading that shortly. And please, let us know what you think. And leave comments. Kamil, it's been great talking to you and to our listeners, have a great day.

This transcript has been edited for clarity.

How can asset managers navigate today’s rapidly changing technological landscape and gain market share despite economic uncertainty? The generative artificial intelligence (AI) revolution is well underway and is already transforming the way asset and wealth managers operate and are delivering significant efficiencies.

In this special edition of our Oliver Wyman podcast series, Joshua Zwick and Kamil Kaczmarski, partners in our Insurance and Asset Management Practice explore The Generative AI Tipping Point, providing analysis and interviews with senior leaders from our global wealth and asset management report. Joshua and Kamil discuss the asset management industry’s outlook and specific actions to capitalize on opportunities. They share the revolutionary power of generative AI and how it can be implemented to drive better investment decisions, engage clients more effectively, and enhance productivity across the value chain.

In this podcast, Joshua and Kamil discuss:

• The state of the global asset management industry at this pivotal moment.
• The inflection point we call ‘the big sort,’ and what it will take for organizations to land on the right side of this reshuffling in the industry.
• Specific actions to reimagine their operating models and build greater resiliency in their businesses.
• The levers management can pull to drive profitable growth and operational resiliency.
• What makes this moment a tipping point for generative AI and how generative AI is already impacting business and society.
• How and where asset management leaders might deploy generative AI to reduce costs and gain competitive advantage.
• Unlocking the power of generative AI for building talent, processes, and business models, and how teams can begin to access its impact.

Joshua Zwick: Hello, and welcome to The Generative AI Tipping Point, a special edition of our Oliver Wyman podcast series. Today, we're going to dive into the global asset management outlook: What's trending in the industry and how managers can win market share and drive growth and build operational resiliency, and then we’re going to pivot to the hot topic of the year—generative AI—and what it means for the asset management industry.

I’m Joshua Zwick, a partner at Oliver Wyman who focuses on the asset management industry. I'm joined today by my colleague, Kamil Kaczmarski, a partner from across the globe based in our Frankfurt office.

Kamil, are ready to talk about the report, The Generative AI Tipping Point, and all the interesting findings that we have written about?

Kamil Kaczmarski: Absolutely. 2022 represented a wake-up call for asset managers. After years of buyout that allowed us to hide underlying fragilities, 2022 highlighted how sensitive asset managers are to market downturns, especially on the traditional active side.

The beginning of 2023 was a bit of a recovery and a rebound in the market, so it helped to ease the sting a bit, but the future environment still looks pretty much challenging. We have uncomfortable high inflation and still a lot of geopolitical tension. So, all of this is actually fueling instability and a little bit of uncertainty about the outlook for growth and especially wealth creation.

I think right now we are in an inflection point, an inflection point that we call the ‘big sort’ for asset managers. There will be a reshuffling of priorities, but also a shuffling of competitive forces. These conditions will exaggerate the differences for managers that can navigate these conditions and be on the right side of the ‘big sort’ and those who can't. And actually, those who can’t, will find themselves in an unhappy position.

So, the question for asset managers right now is, ‘what do I have to do to be on the right side of the big sort?’ What are the winning strategies from those changes in the market? This is what we have talked about in our global wealth and asset management report, and we will discuss today in this episode.

Josh, if you look back at the findings of this report, what were the biggest surprises for you?

Joshua: I think the first thing that stands out is how generative AI is no longer on the whiteboard. It's not just a bunch of people standing around talking about ideas. It's actually being used in business. It's coming out of the laboratory into the production lines and that's creating real impact. We'll talk about this today, it's really much more in the front office and in client-facing areas — where it's having the biggest impact. And we're seeing huge impacts in the order of 20% or 30% in terms of efficiency gains and in particular other areas where it's even significantly higher. So that's going to be a big area and something that is moving really, really, fast.

The second thing that was surprising is the huge opportunity in traditional active. We all know that the secular dynamics of the flows going from traditional active to passive that's still continuing at a pace. That's not slowing down. The analysis we did, which we updated from a couple years ago, shows the trend is still very strong if not stronger, and that the actual flow between active funds, from active-to-active funds, is more than three times higher than the flow from active to passive funds.

So, what does that mean? What it means is that there's actually a tremendous amount of money up for grabs. Money in motion if you will. So, while the pie might be shrinking, if you can get a larger slice of that pie, it's actually a huge opportunity, particularly when we think about kind of the revenue weighted perspective because passive, as we all know, there's not that much money to be made unless there's massive scale there.

Kamil: I love this analysis because I think for a long time there was the ongoing debate about whether active is dead. I think if you look at the analysis and the numbers, it’s positive news and there is still enough revenue to fight for. But Josh, if you think about the kind of levers those managers can pull, what will come to your mind?

Joshua: Yeah, of course, if you have a great performing fund or fund lineup, and you have in-demand strategies, that's always going to be a great thing, right? But as we've seen time and time again on average, we don't see the active, at least not in the traditionally active world, generating significant performance above benchmark.

Obviously, everyone wants to have better performance. But there are other levers that that you can pull and three came out in the research. The first is around product innovation. When you actually look at the data, what we found is that newer funds in terms of the flows that they attract are significantly higher and do significantly better than older funds. I know there are some structural reasons for that. There are marketing dollars put against newer funds. Those newer funds ideally would address a particular market demand at that time. Structurally, they don't have long periods of underperformance where people will be redeeming. But with all that said, what we're seeing is a real advantage to those organizations that can actually put good product out there that meets the demands of the time. And so, I think that's one.

The second was around distribution in service. And we talked about distribution alpha, service alpha—this is real thing. So, from the data we looked at a wide variety of strategies and grouped them by performance and we actually zeroed in on those that did the worst, right? The bad performing strategies across a large number of managers. And then we classified those into, well, if you had bad performing strategies, what's the difference between those managers that did well in terms of retaining AUM or actually driving additional flow versus those that didn’t? We saw a huge skew between those that had really good flows despite the bad performance relative to those that didn't. And you'll see strong distribution networks obviously make a difference. The service levels that are being able to provide it, whether it be solutions and advisory capabilities that are kind of overlaid in terms of the product, that creates a degree of indispensability for these managers to their clients and that shows up in the data.

The last bit of analysis that I thought was interesting as well — you can call these all different forms of alpha, there's performance alpha, product innovation alpha, there's distribution in service alpha. And the last bit of additional non-performance alpha that we saw was around fee structures. And this was also interesting. The funds that had lower fees garnered a significantly greater share of flow. So, what this is suggesting is that investors like the promise or prefer the promise of lower fees relative to the uncertain promise of outperformance at some future point in time. So, being really strategic about how you think about fees makes a big difference because even if you have bad performers, those that had lower fees typically did much better across the dimensions around collecting AUM and retaining assets.

Kamil: Super interesting.

Joshua: It is. So yes, there's probably one or two other things I thought stood out, Kamil. One was actually around pricing, particularly institutional pricing. And we think that managers are leaving a lot of revenue on the table. And so again, we try to let the numbers do the talking here. And we know that published fee rates are significantly higher than what ultimately organizations end up paying. What we found here is that we have some information that shows there are massive skews in terms of the price paid for the same mandate, for the same size of that mandate, across the client base. So, you look at the skew and see some have paid significantly more or significantly less for the same product at the same size level.

If you narrow that band and be more selective in terms of how you manage the discount and pricing process, we see a tremendous amount of money that's being left on the table. But we do think that this is a lever that organizations or asset managers can pull. Being much more disciplined how they think about pricing strategy around client value. How they utilize data to really understand what the long-term client value is, right? How the incentives are aligned and avoiding a situation where salespeople will give away anything at the point of sale to get that gross revenue in the door and think much more strategically about how to price the product, with the broader context of the relationship obviously and the kind of long-term value of that client. And that’s not an easy thing, right? It’s not an easy thing to do. But what we’ve seen for those that can do it quite well, there’s a significant advantage in revenue uplift that you can get from kind of more effective and a negotiation tactics.

Kamil: Interesting. What will be your final surprise?

Joshua: The final surprise is, as you know, we've talked for a long time about the importance of and the growth engines of private markets and an alternative to some extent, and we still see this. That's going to be a key driver of growth going forward.

And we've seen a lot of activity from traditional managers looking to get into this game in many cases inorganically. And I think while that can provide a lot of benefits and tap into a growing market, we have also seen huge operational challenges associated with that. So, if you actually look at the benefits of bolting on a third-party, an inorganic kind of private market strategies or teams or business into the kind of the core franchise, what we see is there can be a lot of friction, whether that is kind of operational-oriented friction, whether it is going to challenge in terms of kind of unifying platforms either from a technology perspective or distribution perspective. That can actually weigh on the overall kind of operational efficiency of the business. And so, the point being, it's not one of those things where you can just kind of bolt it on and everything's all good. It takes a lot of work to successfully integrate and actually achieve the benefits that are the basis for those types of strategic acquisitions.

So, I’m interested, Kamil, in terms of some of the things that that you might have taken from all the work we did together.

Kamil: I think I’ll highlight two topics. Connected to what you just talked about with traditional players expanding to alternatives, what we said at the beginning about the ‘big sort,’ I would expect, some sorting—rebalancing—within private markets. So, it’s definitely one of the key growth zones.

But if you double click private markets and then look at what's happening underneath, there’s some very interesting dynamics. If you think about real estate, this is an asset class where there's still a bit of concern in the market. With this asset class, it's still a rebalancing of the attractiveness, relative attractiveness of real estate.

Private debt is definitely right now people's favorite, I remember one of the executive interviews when a person said it's like Christmas for us. It was a private debt player. Let's see how long, but it is definitely an area where we see not only in corporate private, but also in real estate where the yields are so attractive versus equity, real estate equity.

There’s a lot happening where those who didn't have experience yet are entering this space and others are simply realizing, especially right now in an inflationary environment, it is actually a very attractive asset class.

And then, finally, two of my favorites: infrastructure and natural capital. I think both of them are strongly benefiting from the entire trend we have right now in responsible investing and especially as an asset owner, are finding suitable investments that will help achieve ambitious targets. Many are pushing on the boundaries of net zero. You will need a lot of support from investing into infrastructure and especially natural capital such as forestry, to reach those ambitious targets.

So, as depicted in one of our charts we show how the revenue margin changed over time for the industry and the cost margin. There was a very clear drop on the revenue side. But the cost side remained stubborn. There was little change on the cost side. And this is the over all problem we as an industry are facing — that we are not yet entirely prepared for a potential market downturn.

I think some of the players were stuck in the middle, especially where they probably lack the scale. They will be hit hard if we enter a market downturn that will last more than simply just one quarter.

So, what we did in this report and what I really like is our menu of different levers that asset managers can pull. It’s a healthy check to go through them and understand where you are. And if you just look at them, there are some that many managers have addressed or are looking into. These are the typical cost ranges I would say 5, 10, 15, up to 20% of your cost base. But if you really want to think hard about the potential of really reducing your cost base beyond the 20%, this is where you have to make some really tough choices. Tough choices about your structural cost base. And tough choices about questions, such as your location footprint. Do I have to be present in all those locations and the satellite offices? Tough questions about your investment landscape or the capability landscape. Is it credible for me to, for instance, manage Japanese Yen out of Europe?

I think the industry is right now at this inflection point, but those hard choices are on the table. There's an scenario out there of a longer downturn and we will have to operate in an environment where the cost has to be adjusted.

Joshua: Yeah, and these are perhaps somewhat provocative statements and the possibility to get 5% to 15% or around 20%, but there's a lot more on the table if managers are really willing to make those hard decisions.

And so, one of those things that we’ve talked about, obviously a lot in the report and I'm talking about now, is what's the role of generative AI in this? I thought, Kamil, I'd walk the audience here through some of our perspectives on Gen AI that we covered in the report, and a keyinitiative that’s possible is changing the operating model to drive significant efficiencies. But that's just part of the story, right?

So perhaps, if you don't mind, we’ll shift over to the Gen AI side of the discussion.

Kamil: Let's do that.

Joshua: Alright. So, one of the things we've coined is the term the ‘big sort,’ and hopefully that is evocative of the way that we see the industry shaping out. The other thing is where we are as an industry and this concept around the generative AI tipping point, and why this time is different.

And we see the generative AI tipping point being driven by three things. There's three aspects. The first is the higher degree of accuracy. And you know, if you looked at how GPT-1 or GPT-2just responding to a simple email, what resulted was pretty bad. It's clearly not coming from a human being. So, if you are using that as a chatbot it is not good. If you look to take an exam, like say the Bar Exam, I believe GPT-3 passed it by 10%, but GPT-4 passed it by 90%.

Greg Jensen, co-Chief Investment Officer of Bridgewater Associates was on a podcast recently and he commented that GPT-3.5 was able to pass the investment associate exam given to all investment professionals at a 20% level. With GPT-4, guess what? 80%. Having an infinitely scalable analyst similar to those from the top universities at your fingertips is pretty remarkable, right? So that accuracy is just crazy.

Second is the broad range of the application, the generative AI applications, nowadays. And it’s not just about generating text and creating new ideas. It's also obviously creating new images or new music, but I think what's remarkable is the generality of it. It can be applied to revolutionize science, right, and being able to review journals on protein folding and come up with novel new molecules, right? Or review massive amounts of investment research. Or it can just be used to create a really clever holiday card, right? It is amazing how broadly applicable these types of applications are.

I think the other thing, and this is probably the most transparent or salient for most people, is the accessibility and user friendliness of interfaces. Anybody with a high-speed internet connection and maybe a couple bucks in their pocket to go to GPT-4, you have complete access to absolutely groundbreaking technology, right?

So, you combine those things — higher degree of accuracy, the broad range of kind of applicable use cases and the accessibility, the user friendliness — it really is a game changer.

Kamil: And within the asset management industry, if you think about the value chain, what do you think would be the largest opportunities?

Joshua: Yeah, so based on our research and talking to a large number of asset managers in this space, I think the first point, I alluded to this earlier on, is that the revolution is well underway. These are not use cases where they're not just on some kind of paper or on a whiteboard. Organizations are actually doing. And what we see organizations actually doing is focusing much more on the frontend where there is client interaction, where there is the investment process to some extent — I'll clarify that in a second — and areas where there are huge advantages that the nature of Gen AI can actually create. I think this is important point.

So, before I get into the specifics, I just want to make a one comment which is that Gen AI is not perfect for everything. It’s not the panacea, right? There are certain types of things it does really well, certain types of things where other types of technologies or process redesign or other types of machine learning models are much better for. But when you have a situation where there's some degree of routinization, something that requires the creation of new content and to some extent customized content, and to the extent that there's a large amount of information to synthesize, if you have those conditions, Gen AI is a really good tool to utilize in those situations to generate massive efficiency gains.

And so where we see those types of situations being most prevalent, it's often in the front office, client facing, to some extent the investment research areas. And so, think about writing, transcribing, and synthesizing calls. There's really no need for human to do all of that, right? Think about spending time drafting insights about clients or talking points to support targeted sales. There's really no reason for human beings to do all of that? Think about drafting RFP responses or customizing marketing materials or thinking about synthesizing research reports, right? Or reviewing and summarizing mountains of due diligence documents as part of a deal. Yes, you still need humans in a weigh in, and we'll get to that, but a tremendous amount of progress or efficiency gains can be made by not starting with a blank sheet of paper. Imagine we've all been in a situation where we're looking at a sheet of paper or a mountain of documents to synthesize and go, where do I even start? Gen AI can actually get you way down the field to use a football analogy, sorry, Kamil, and to the 50-yard line and well into the end zone in some of these cases, right? And so that sort of uplift, we think is tremendous. Now, we don't see it being used anytime soon for thing such as picking stocks.

Kamil: That will be my next question. It will be my next question because if you would tell me right now that you've found the tool that will tell you what stocks to pick or how to push in the market, well, Josh, believe me, I think we would be in the Bahamas sipping pina coladas, but tell me more.

Joshua: I would agree with you, but Gen AI is say, 80% right in some cases, much less in other cases, right? We’ll talk about the limitations shortly. And what often generates alpha is the last 1%. And so, the generalized answers that it provides are obviously not overly useful for that. But for developing new ideas, new investment concepts, identifying potential risks, things you may have otherwise missed just because of the capacity constraint. You can read and research so much.

Those are the types of things that can be supportive of an overall investment process. And so, it's not going to be used,to trade, build entire trading systems off of it. But it can be really valuable as inputs into the overall process. And that's really kind of what we're seeing and where you think the greatest benefits are likely to be.

Kamil: And if you think about the kind of elements or prerequisites to be successful, is there anything to highlight? Is it for everybody or are there any elements you simply need to have in place to really harvest this potential?

Joshua: Yeah, I know that's a good point. And there's no doubt there's a degree of democratization that this technology provides. I can actually now write code. As long as I know that the questions I want to ask, and I have some understanding of what I'm trying to achieve I actually don't need to be that skilled in C++ or Python to actually write the code.

And so, there's an element of this that actually democratizes things. Any time you democratize something in an industry that is largely commoditized and is looking for that small edge to generate alpha, something that's democratized is actually not all that useful. And so, while it can provide a lot of these efficiency gains that we're talking about, it really just kind of raises the bar. Everyone’s going to have to do that. And I mentioned the speed, this is moving fast.

And there’s a couple of success factors that we call out the paper and I'll just kind of highlight here. One is the proprietary data. We're not going to have institutions creating their own large language models (LLMs) and training these models. It's just way too expensive and way too complicated. But to the extent that they can be fine-tuned to capture the tonality of a particular way in which an organization communicates to capture all the insights from all the emails, the research in the reports, the market commentaries, and the client dialogues, all of those things, those sort of insights, those nuggets can be harvested and combined with the capabilities that these large language models have, by being trained on basically all of humankind’s knowledge.

It's actually allows you to do much more interesting things that are really specific and harness specific kinds of capabilities, insights, research, , if you will, of these organizations. So, in a way it becomes less of a data science question and in many cases more of a data engineering question: How do I get all this data together to basically tune these models and incorporate these sorts of insights into something that is really useful and arguably differentiated relative to all others?

Another key differentiator, Kamil, is probably more what we like to call a systems-based approach. And I think while there's a tendency to write down a lot of use cases and we've been with some clients who have probably hundreds of potential use cases. How do you think about them more holistically? What's a strategy that can be applied to really focus on the on the return on investment (ROI), because that's another element of this, which a lot of people are very excited about the prospects. And hopefully, from my voice you can detect how excited I am! Inside Oliver Wyman we've created some applications that I'm currently using that are making my life significantly easier, including some of the more administrative tasks.

So, I'm really excited about Gen AI, but we need to think strategically about what the return on investment (ROI) actually is because there is a cost. And in many case the cost isn't all that high, but it is high in terms of people's time. And so being clear in terms of what's the overall strategy, how do we think about this more systematically across the organization? How do we allocate resource in the best possible way? That can make a huge difference. And I think that can create some sort of systemic or systematic advantage over time.

The last thing that is potentially a differentiator in terms of how you can deploy these capabilities is reimagining the operating model. And one of the big changes you often hear in these types of conversations around the applications of Gen AI is that it changes human beings from creators to validators, or to reviewers. And it's not to say — we still need to create good content, but ultimately the models need to be trained and as an industry, we're facing some of these challenges at the moment. But being able to recognize the changes that these sort of models and technologies create. If we are doing much more reviewing, does that mean we need to have a center of excellence, in terms of reviewing things that go out to clients. How do we ensure that we keep humans and the right humans are in the loop because we're probably not at a situation yet where you just literally are letting one of these AIs go into the wild and actually provide advice, for example, or detailed information to end clients.

Organizations need to understand how the roles will change and how to centralize things. Do we need to decentralize some things? Do we need to change our processes? Those can create sustainable competitive advantages, but it really requires understanding how the technology impacts the way the organization operates.

And the last I'll point, and there are others, but this is around the data governance and controls. Organizations have information that's sensitive. Secrets for their own kind of trading strategies. Other organizations obviously will have private information about individuals, whether the employees or clients. And letting that of seep out is a real problem. And this creates a big perimeter or surface area for malicious actors. And so, I think those organizations that take a really sound approach to managing the data governance and the data security are going to be essential to making responsible use of these technologies.

Kamil: Super interesting. So, Josh tell me, what do you think will keep our clients busy for the next couple of months when it comes to Gen AI.

Joshua: One is around this systems-based approach to thinking about their generative AI roadmaps. And so, we see a lot of organizations putting a lot of ideas out there. But how do you actually prioritize them? What sort of frameworks do you want to use? What are the criteria for where this is a good idea versus where it's not? Where is the ROI actually worth it? And so, a lot of organizations are struggling quite honestly with the prioritization and how to create more of a systems-based approach to deploying this type of technology at scale throughout the organization.

The other area is around execution implementation. Like I said, this out of the lab and it's into the system. It's into production right now. And so, we spent a lot of time developing tools internally to support our business and our ability to support our clients, but helping our clients actually develop their own tools, host their own applications, and actually train in a way that allows the organization to generate proprietary insights and something that’s differentiated that only that organization can do relative to all others.

So really that execution and implementation, that's where organizations are moving to and that's the types of calls that we're receiving.

Kamil: Excellent. Josh, we are about at the half hour. Any final words of wisdom as a closing remark?

Joshua: Well, I've been doing a lot of talking, Kamil. I'm tired of listening to my voice, so maybe I will put you on the spot. Anything else that you wanted perhaps to comment that we’ve missed or any closing thoughts for our audience who made it this far with us.

Kamil: I think I would encourage everybody to try and pilot the potential of generative AI in their organization. You can start small, pick an area where you understand there is a lot of manual intervention. There's a lot of creative work that's being done. And as you know, my history before I joined Oliver Wyman, I started my career in product development. And I have to be honest that I caught up with some colleagues from a different company and they told me how they do the process today and it is still the same. I think Gen AI gives another opportunity to really rethink how you get things done, especially the area, as an example, product development, right? When it comes to writing a narrative, a sales narrative. Or even the repetitive process of launching a product where very often you have to go through internal committees, and later you have to go to the regulators, and every time you just need to compile various documents or text Gen AI can help you create efficiencies. So, I would encourage organizations to build use cases and try it out. Because the potential, once you see it, is there already. I think it's eye opening. To see how much potential that we have and how much it will keep us busy for the next couple of years.

Joshua: Well, I think that is exactly right. I actually love playing with AI. As long as the organization keeps in mind the things we’ve discussed and considers how to make it successful at scale. That’s where I think those are going to be on the right side of the ‘big sort’ versus those on the wrong side.

So, with that, I hope everyone enjoyed our conversation today. I've been involved with our Global Wealth And the Asset Management report for many years now, I think this really is the best one we've done, so I encourage all our listeners to read through it and hopefully they will gain unique insights out of it.

We're also doing another podcast on the wealth management side of report. So, my colleagues will be leading that shortly. And please, let us know what you think. And leave comments. Kamil, it's been great talking to you and to our listeners, have a great day.

This transcript has been edited for clarity.