- 4 minutes ago
After years of record AI investment, PwC's global research reveals that 56% of CEOs from the world's largest corporations have seen zero return, on revenue or costs. And yet, financial markets are pricing AI as the defining opportunity of the century. AI will not lift all boats but it seems a lot of people missed something or skipped the fundamentals? Solving this execution gap will define who wins and who loses.
Category
🤖
TechTranscript
00:22Hello everyone, hello, Valerie Bodson, CEO of Amundi, and Mohamed Kande, global chairman
00:29of PwC. Thank you very much to be with us. Are we missing something? Apparently, CEOs
00:35are very disappointed about the result of AI in the organization. You said that PwC
00:43surveys say that more than half of the global CEOs don't see any result, or very few of
00:49them. So maybe we are going to try to know if we are missing something. When you see
00:55the market value of the AI companies in the world, apparently there is a big thing going
01:02on. So why can't we take advantages of all that new technology? So maybe to start with,
01:11let's see how in your own organization, in your own company, you implement AI, and what
01:17kind of advantages you were able to be sure. Maybe we can start with you, Valerie.
01:24Well, thank you very much. Well, first of all, I will try to give you some positive examples,
01:31because there are some companies who benefit from AI, and Amundi starts clearly benefiting
01:35from AI. So just to give you a rough idea, Amundi is the largest European asset manager.
01:41We manage 2.4 trillion in 35 countries with 6,000 employees. So that's where we try to
01:48work on AI. Let me first start by a conviction. AI represents many opportunities. It also represents
01:58some risks. So you absolutely have to implement it and to adopt it with ambition, but also in
02:08a thoughtful way, if I had to summarize. So to answer your question, what have we done? We have
02:14started in 2022 by building strong data management and building a platform on it called Alto Studio. It is
02:25a proprietary platform. It is a secure platform, a multi-LLM platform for all our employees that we
02:31spread around the company in 2025. To give you an example about the efficiency. In 2025, 20% of our
02:42employees were using it more than once a week. Nine months later, more than 50% of our employees use
02:50it at least daily. So this is an evidence, obviously, of the gains in efficiency and productivity. And we're
02:58also working, to give you another example. And we have worked and launched 20 new applications dedicated
03:04to one team or to one activity, where we gain efficiency, such as client servicing. Typically,
03:11the 4 million retail clients we are covering on employee savings scheme. We have gained at least 20% of
03:20efficiency in managing their questions and managing efficiently our answers. I could speak about the RFPs,
03:29more than 2,000 we are having per year, where we also gain very clearly 50% of efficiency. And
03:37last but not least,
03:38I wanted to give you for financial companies something which is very important. We have a lot of controls to
03:44implement. And we manage to do now in minutes what we used to do in ours. So there is efficiency
03:53coming step by step, thanks to AI. Mohamed, 360,000 employees in the world?
04:01Something like that? Yes, 367,000. Okay. How did you manage to make them use AI? And what are the
04:09rules now?
04:10A couple of things that we did originally, a few years ago, when we started seeing all the different
04:17tools coming out, whether it was chat GPT or, you know, co-pilot and you name it, right? All of
04:22the
04:22different large language models. The first thing that we did was to democratize access to all the tools
04:28when it comes to AI across all of PwC. Why? Because we did not want people to fear artificial intelligence.
04:35And people fear what they don't understand. The first thing we did was to give them access to the
04:42different tools. And by the way, we all know in the room that AI is not free. So we needed
04:47to make
04:47the capital investment to make sure that all of our people weren't exposed to artificial intelligence.
04:52The second thing we did, and which leads to the first question you asked, what are we missing?
04:58We found out that it was important to do experimentation at the beginning,
05:02but experimentation needed to move to scaling. Not the use cases, not working in the lab,
05:08but how do we scale the use of artificial intelligence from across the business? That's
05:14when we get a lot of the different adoption. The third one, when we think about artificial intelligence,
05:20and you mentioned it, originally we looked for productivity improvement, like many other organizations,
05:26by the way. But now we are finding out that it needs to be about growth of additional values that
05:31created, like you're doing your customer service, for example. And that value creation and growth
05:36became more important than productivity. Because productivity was looking at you taking something
05:41away from people. Value creation, people could give more using artificial intelligence. And the last
05:47point I will mention is the fact that we also realized that artificial intelligence was not only about
05:52the technology. It was about the business transformation, meaning that not just doing the same thing using artificial intelligence,
06:00but doing things differently, reinventing the business itself, so that we can get actually more out of artificial intelligence.
06:06You speak a lot about people. Do you think there is a risk of two categories of jobs and job
06:11market?
06:11You mean the one with routine jobs, who are very affected by AI, and the other with maybe augmented jobs
06:19with AI. Do you think this is the risk of the two parallel job market?
06:23I would think that, I might even say that we'll have three different markets. The first one would be routine
06:28jobs that are being impacted today by artificial intelligence, where people get to do, by the way, a lot of
06:35the people doing the routine jobs don't like the routine jobs.
06:37So artificial intelligence, actually giving them something more for them to be able to do the jobs differently. Other jobs
06:45will be augmented, because you get more productivity and more value out of employees by using artificial intelligence.
06:52But there is a third category of jobs that have been created, jobs that don't exist today at scale, because
06:57of artificial intelligence. We just don't have enough AI engineers in the world.
07:01So all that new jobs will be created like we had 25 years ago with the internet, where you had
07:07new job categories, right? So we also have to understand that artificial intelligence is not just going to improve what
07:14we do today. It's going to create something new after that.
07:18Maybe let's speak about growth on revenues. You said how, you know, you use the AI tools to gain productivity,
07:25and that the result was good, but not as spectacular that they could be. But maybe on revenue and growth,
07:32new services, new opportunities, AI can help you.
07:36Maybe it's what we were missing. I mean, there's 55% of CEOs who don't see the gain. Maybe they
07:42should look more about what new revenue and new activities they can make with AI.
07:48Well, for the time being, I mean, results are exactly as I was expecting them to be. As I mentioned
07:55in the introduction, we want to adopt it with ambition, but in a thoughtful way, at the right pace, at
08:01the right reason.
08:02I think it's really important, as you were mentioning, Mohamed, to manage the adoption of the tool by our employees.
08:11It needs to be done with the people.
08:15And as you were very rightly mentioning as well, using it to generate more revenues is obviously a very positive
08:23way to sell this tool to our teams and to our clients.
08:29So what did we do in terms of revenue? First of all, of course, we always wanted to remain to
08:37have proprietary tools because we consider that owning our technology
08:41technology allows us to be more innovative. It accelerates innovation. It has always been true and it is true with
08:49AI.
08:50Second, as I was mentioning, it helps us to be more efficient with our clients, especially in servicing them. And
08:57the better you are with your clients, the more revenues you make.
09:00And third, this is probably the most important part. At Amundi, we have launched in 2020 and it was not
09:07linked to AI at that time, a new technology business line.
09:12As we own our technology, we decided not only to provide asset management, but also to provide and to sell
09:19our tools to our clients.
09:22And that's where it gets interesting in terms of revenues, because the famous platform I was mentioning, Alto Studio, we
09:30now sell it to our clients as well.
09:33So this is pure revenues first. And second, one of the areas where efficiency is the most important when you
09:42use AI is IT development.
09:44This is known and proven everywhere. And we can see it at Amundi. Nearly 20% of our staff are
09:54in IT and technology.
09:57And we see how by transforming the way they're working with AI, how we can be more efficient.
10:04So it's also a source of revenues because it allows us to provide our systems more efficiently and more quickly
10:11to our clients.
10:13Mohamed, when you work with your customer in 135 countries around the world, what is the part of your activity
10:21in developing productivity on one side and increasing their revenue and their growth with new tools, with new AI tools
10:30on the other hand?
10:33It's about 50-50 when it comes to the focus on productivity, but also on growth.
10:39But as Valerian mentioned, you also have the management of the environment, the controls that need to be put in
10:46place.
10:46So we spend a lot of work making sure that AI is adopted responsibly.
10:52And to make sure that this happens, you need to put risk management frameworks so that you don't have hallucinations,
10:56for example.
10:57The data management infrastructure needs to be set up in a way that the data resides within the different organizations.
11:04Governance about what AI has access to versus not access to.
11:09And the decisions that need to be delegated to artificial intelligence.
11:12So that part of the job, I would say, is the first thing to do.
11:16Any organization that wants to adopt artificial intelligence at scale, not just for experimentation, needs to do the basics around
11:24governance, around data, just to make sure that they will have responsible use of artificial intelligence.
11:30But now we're seeing that the demand is both productivity improvement, but also growth and value creation, including innovation, by
11:37the way. Absolutely.
11:39We talked a little bit earlier before entering the scene, and you were telling me that the cultural approach of
11:46AI is very different from a region to another.
11:49Could you elaborate on that?
11:50So, there is a perception in some parts of the world today that artificial intelligence is a bad thing.
11:57You hear in the press.
11:58You hear the pushback around data centers, around AI, you know, taking jobs away, et cetera.
12:04And you see it mostly in the, for lack of a better term, the Western world, right, where jobs might
12:10be disappearing.
12:11But one of the things that we just discussed is that we should not project any feeling or thinking about
12:18artificial intelligence from one part of the world to another one.
12:21Because when you find yourself in Asia today, for example, whether you're in Japan, in Korea, even in China, and
12:28also in the global south today, the value of artificial intelligence is different.
12:33Why? Because it gives access to more resources to people.
12:36So, it is not the same feeling.
12:39So, we have to understand that the consequences or the implication of having access to artificial intelligence is very, very
12:47different depending on where we find ourselves in the world.
12:50And the mistake for any executive to make is to assume that everybody will react the same way.
12:55It is actually not the case.
12:57There is a level of sophistication now that is needed to drive the adoption of AI in different parts of
13:03the world.
13:03But there is a huge amount of money being invested in the AI technology right now.
13:10Is it worth it?
13:11I mean, how are we going to control the cost?
13:15We don't know how, you know, the LLM will be built in the next years.
13:20I mean, for now, I think it's really, it's a very bargaining price.
13:25But what about the, you know, the tokenization of the market or the billing market or that?
13:29Are you worried about your, you know, the cost of all that?
13:32It could be very, very huge in a few years.
13:36Yeah, absolutely.
13:37I mean, I'm absolutely certain that it is the most important part and point that all CEOs of the world
13:46must monitor.
13:48It's the cost of use, as I was mentioning in my introduction.
13:52It is the main risk which goes with these large opportunities.
13:59So how do we do it? How do I think about it?
14:03Well, first of all, like all providers, you need to be a multi LLM.
14:08You know dependence on one provider only.
14:11It's a basic, but it's true in artificial intelligence as well.
14:16Second, you need to make sure in large organizations like ours to give access in a secured way to the
14:25right model.
14:26You don't need to have the best and most perfect model for basic tasks.
14:35It costs a lot of money for nothing.
14:37So it's also a question of education and training your employees to use it cleverly.
14:47It's about prompt engineering that you need to learn as well.
14:50So it's a lot of education, which is done in a very modern way.
14:57At Amundi, we have champions in all the teams.
15:00They share with their colleagues how to do it cleverly.
15:03And it's done with a lot of interest.
15:05And last but not least, I think like in any well-governed organization, you need to centralize absolutely the cost.
15:14And you need to monitor it on a daily basis and on a monthly basis.
15:19And you need to cut it when it goes too high.
15:23As simply as that.
15:24In that case, you can monitor costs.
15:27Mohamed, how can Europe, you know, go through that?
15:30Do you see you have a worldwide vision on the whole?
15:33You just told us how AI is implemented everywhere.
15:37There is a risk of a transfer of added value from, you know, certain regions to others, from Europe to
15:43the US.
15:43I don't know.
15:44I don't know who, you know, you will get the revenue from the job and the work made by AI
15:51and not any more humans.
15:53How do you see those huge transfer of value through the world?
15:57So, a couple of things when you think about the value of artificial intelligence.
16:02A lot of the time when you think about artificial intelligence, we think about the models, right?
16:07But now we are thinking more about the ecosystems.
16:10So, for example, we talk about the jobs.
16:12AI, on the domain side of the use of AI, there is implication of jobs and job creation, et cetera.
16:18But on the supply side of AI, it's not just about the models.
16:22It's about the chips.
16:23It's about the data centers.
16:24It's about energy, industrial equipment.
16:27Look at all the other industries that are now creating jobs because of the demand on the AI front.
16:33So, it's a good news story when you think about this.
16:36So, when I'm thinking about Europe, even if Europe might not be building as many large language models like you
16:42have in many different countries,
16:44there are a lot of, in the ecosystem of artificial intelligence, it's a big play for Europe around industrial equipment,
16:51around energy.
16:51Look at, in Europe, the expertise around nuclear energy resides in Europe, actually here in France.
16:57We all know that.
16:58So, Europe has a lot to offer on the supply side of artificial intelligence.
17:03That's the first thing.
17:04But now, when we think about the use of AI, what we have seen, technology can be invented in one
17:12part of the world.
17:14It is also very important to consider the diffusion of the technology.
17:19You don't have to invent the technology to be the recipient of the benefits of the technology for diffusion.
17:26And that's what I see, honestly, for Europe to see.
17:28Europe will have the ability because of the ecosystem, the people, the environment, the policies, and also the talent to
17:37have accelerated diffusion of artificial intelligence.
17:41Valérie Amundi is one of the world, you know, leaders in asset management.
17:46You have huge money to invest.
17:49What could be your role also in helping Europe to gain and to keep her sovereignty in those new AI
17:58worlds?
17:58Well, I think it's a very positive move we've seen since four or five years.
18:05Europe is realizing that sovereignty is important in everything.
18:09It's important in energy.
18:11It's important, of course, in food, health, defense, but it's also important in finance and in technology.
18:20So, I think the trend is a good one.
18:23We need to go on pushing it as much as possible.
18:27We need to support our European champions, Mistral being one of them.
18:33We need to go on working on the ecosystem, as you were very rightly speaking, because we had a lot
18:41of strengths.
18:43We need to work on a European cloud.
18:45I think it's one of the priorities we have today.
18:48And as very large financial investors, we very much encourage our clients to invest in Europe through diverse regulation, but
19:05also because we believe in it.
19:07We believe in sovereignty.
19:08We launched, for instance, a fund three years ago called Europe and sovereignty, where we invest in all the companies,
19:16including tech companies, which are supporting the independence of the technology of Europe.
19:23Well, I mean, AI is like a tsunami.
19:25You know, we see it approaching for every sector of the economy.
19:31What advice would you give to each of the companies to be on the winning side and not the losing
19:39side?
19:39Is it investing more in AI, investing in human?
19:43What is the key now?
19:46I would have, I think she very stated, right?
19:49The key is to aid and don't lock yourself into one solution.
19:53You need optionality when it comes to artificial intelligence.
19:56And the second thing I will say is to continue to learn.
20:00Assumption that we are making about the possibilities around artificial intelligence, they change every six months.
20:06So now there is a responsibility for executives to make sure that they test the assumptions all the time and
20:12they keep asking questions about what could I do differently.
20:15It is a very, very different world with artificial intelligence because when things change every six months, every three months,
20:21we have to constantly reassess the decision that we're making without changing our overall strategies.
20:28It is not like it was in the past saying that it was all about scaling businesses.
20:32No, this is a game of innovation.
20:35We believe that a lot of companies that will do well in the market will be the companies that are
20:41very good at learning about what's happening and implementing it within their businesses.
20:45It's a different job.
20:47Your job will be different in a few years, you think, in this industry, in your industry?
20:52You mean my own job?
20:55My own job is changing every day as yours, I assume, so that's the beauty of the job.
21:00No, I think for the CEOs in the current environment, which is honestly fascinating, what is really important is to
21:10embrace these changes, to adopt them with ambition, with vision to explain where you're going.
21:22But also to embrace them without precipitation, because we need to learn, and also thoughtfully to make sure that we
21:34do things with responsibility.
21:37So this is already my job today, and I think it will still be my job tomorrow.
21:42Thank you very much to both of you.
21:43Time is up, but thank you for the talk.
Comments