Big banks’ tech leaders lay out the future of AI in finance — and its impact on working in the industry

Welcome to The Finance Files, a newsletter from LinkedIn News bringing you the must-read news, views and conversations about finance, fintech and the economy, from Finance Editor Richard J. Chang. Click 'Subscribe' to join the community. You can check out our previous editions here.

More than 200 professionals in artificial intelligence and financial services spanning banking, credit, capital markets and asset management gathered in New York on Thursday for the Evident AI Symposium, where they shared their future plans with incorporating AI into finance, as well as the challenges the industry could face.

Data leaders at banks shared that in the next few years, AI will change the way people work in finance, but it won’t get to its full capacity without throwing out specific use cases in favor of company-wide frameworks, as well as building customers’ trust with their financial data over time.

Here’s what AI leaders, chief executives and head data scientists in finance shared about how they envision the future of AI in finance:

Finance will adapt with growing use cases of AI, changing the nature of work and how the industry hires talent.

Though some fear that AI could replace jobs in finance, leaders stressed that human capital will become more important than ever and drive demand for labor in the industry. Many professionals will have to adapt to using AI as a productivity tool, they stated, but people will still be at the forefront of running banks.

Emmanuel Roman , CEO of PIMCO: “There are many more things we can do with the same number of people that in itself – should everything else be equal – will be good for the labor market, loosen up the labor market and over time significantly increase productivity. Looking at various innovation cycles, the market always overestimated the impact of radio, electricity and AI. But the way it's transforming our business is fantastic.”

Sid Khosla , Americas banking and capital markets leader, EY: "In many ways these use cases will undergo a Darwinistic journey and will evolve as we mature in the evolution of AI. At the same time, AI won't fully replace the human elements of finance. We'll see the human responsibilities of working in banking and capital markets shift with the growth of AI in the same ways we saw this line of work shift three decades ago with the internet."

H. David Wu , firmwide AI head of product and architecture strategy, Morgan Stanley: “I think most banks have a one-bank, one-firm strategy whereby the capital markets business can connect. There are certainly walls around that. But does AI allow us to revisit some of these things and say, ‘Hey, we can meaningfully and proactively suggest this is the banker you should talk to’ because your client just indicated the move to the next phase of their business’ maturity.”

Dan Jermyn , chief decision scientist, Commonwealth Bank of Australia: “One of the things that is dramatically undervalued in industry is this idea that really everything boils down to people. The people that are developing the capability and for what purpose and, more crucially than anything, the customers you're serving. Doing better and better things for customers turns out to be incredibly powerful for us as an organization.”

Ned Carroll , head of data and automation, PNC: “I wonder, is corporate America creating an environment where we can really enable the new generation of computer scientists and engineers coming in on the one hand? On the other hand, we've anecdotally found that the best productivity we get around code generation, as an example, is with our most seasoned engineers. It opens up broader reskilling of the population in terms of AI in the same way the country went through reskilling with manufacturing.”

Luke Gee , chief analytics and AI officer, TD Bank: “The reality is humans aren’t perfect. As we baseline the human capability, that's what allows us to think about what could be possible with an agent, and therefore can the agent be more consistent. The reality is the agent will be consistently consistent or consistently inconsistent. We have to work out whether there is a save and a benefit to building a system like this, and actually work out whether the human in the loop is qualified at the end to actually check the output.”

Trust is essential for AI’s success with customers. At the core of this is data security, governance and products that help customers.

Finance chiefs said that while financial companies are positioned to invest heavily in AI projects, many are taking a cautious approach to ensure they can create solutions that are using customer data ethically. Without establishing trust, they said, AI in banking will never succeed.

Shruti Patel , chief product officer for the business banking segment, U.S. Bank: “We definitely need to [deploy AI] faster. But, outside of regulation, banks also have this other burden about social inclusion, social benefit, causing no harm, being fair, being transparent, and explainability. These are some really high criteria to take into account. That's where the constant iteration of your models comes in, because the last thing you want is a credit underwriting machine-learning model breaking down and saying no to an entire segment of population because of how you built it based on proprietary data.”

Andrew Reiskind , chief data officer, Mastercard: “As financial institutions, our trade currency really is trust, as opposed to a lot of other industries that are pushing the envelope a lot faster because they're less worried about trust. It's not as big of an issue as if it starts pulling money out of my checking account; that might be a regulatory complaint coming as a result of that. But that's not going to be the case. Banks are coming across as more cautious, but that caution is well deserved in our industry.”

Ned Carroll , head of data and automation, PNC: “I’ve witnessed a lot of focus on trust with data so you can now expand that to different data sets. I think our focus is really more on ‘are we doing right for the customer?’ At the end of the day, if we stay focused on whether the customer trusts the decision that we're enabling and they trust the data, we build transparency to how we're using that data. I think as long as we stay customer- and client-focused, whatever regulatory construct that's put around us will be fine.”

Bing Xiang , head of AI research, Goldman Sachs: “AI will be secure, transparent, safe and fair. But during the deployment stage, we need to ensure we have a well-established governance process in place. We all want a large AI model to be creative, and we don’t want to waste that creativity.”

Sumitra Ganesh , managing director, J.P. Morgan AI Research: “Essentially humans are providing the goal, so the whole agent-based system gets fired off, some human comes and asks a question, sets a goal, maybe even clarifies the goal. We will probably evolve to more proactive systems where agents come up with their own goals. These require a lot of domain knowledge and nuance.”

AI is currently siloed into many use cases. The challenge is bringing all of that together under one model.

AI leaders from top banks shared that many in finance are trying to harness AI in independent use cases and experiment within specific departments. But they said over the next few years, each firm will have to scrap using AI in separate projects and create a more coherent and unified framework for how their companies as a whole will use AI.

Kristin Milchanowski, Ph.D. , chief AI and data officer, BMO Financial Group: “Big data doesn't necessarily mean better anymore. Precision, I think that's one thing that's often missed. When you're trying for an outcome, what level of precision do you really need from these LLMs, because you probably don't need 99% of what you're doing. So you could save a lot of tokenization, you can save a lot of power and energy, and then you can save on perfection.”

Prem Natarajan, PhD , head of enterprise data and AI, Capital One: “If you say ‘this is my framework. Now I want to go find where are the use cases that unlock maximum value,’ your focus will shift away from the specific use cases to creating a platform strategy that allows your entire company to go find the use cases they all want to develop and deploy.”

Sameer Gupta , Americas financial services AI leader, EY: “Critical work of the banks is being done in Excel sheets nowadays that AI could do instead. But we are deliberately not doing that. And I think the reason we aren't doing that is that we want to do it in a way that the adoption has a productivity element to it, has a stickiness to it, and is a safe way to do that without running them off with.”

H. David Wu , firmwide AI head of product and architecture strategy, Morgan Stanley: “The analogy I'll give you is that these AI agents seem like they can cook maybe one dish at a time, where you teach them to use one tool at a time. I think the promise is, can you get the agent to create the whole Thanksgiving feast. It has to learn how to taste the turkey, has to learn how to cut, has to learn how to use the oven, and that requires a lot of coordination and collaboration. It's not quite there yet, but this is moving so fast. Maybe you get a few agents that are more intelligent that can do these complex tasks.”

The amount of trading account assets held by all U.S. banks topped $1 trillion in the third quarter, led by JPMorganChase . The world’s largest bank has $506 billion in trading assets, representing more than half of the total across all banks, according to data from the FDIC on BankRegData.com.

Trading account assets refer to the securities that banks buy and own for the purpose of reselling in the near term to generate profits. Their rise above $1 trillion – which hasn’t been seen since the first quarter of 2008, months before the housing bubble burst and the great financial crisis began – indicates that big banks are holding onto a greater value of securities than usual and anticipating they will be worth a greater amount sometime soon.

JPMorgan’s $506 billion represents 14% of its total assets, the highest ratio for the lender since at least 2014. It’s steep in comparison to rivals Bank of America (4.4%), Citibank (7.8%) and Wells Fargo (3.8%), but is squarely in the double digits alongside Goldman Sachs (16%) and HSBC (13%).

We take a quick dive into a company making a bet that others are running from.

Payments giant Mastercard is aiming to get rid of card numbers and passwords and replace them with numberless cards with biometric passes such as fingerprint and face unlock by 2030, the company announced this month. Mastercard is the first card payments company to lay out plans to phase out card numbers entirely as e-commerce continues to dominate retail.

Phasing out card numbers could mean a revenue boost for the company, whose goal is to decrease “cart abandonment.” Mastercard hopes that by making shopping quicker by not requiring customers to input card numbers or passwords, while maintaining security, they can reduce friction in online sales. The company’s internal research found that many customers didn’t complete purchases because of friction such as not having their cards with them or having to input too much information.

The company has already simplified some purchases in the form of Click to Pay, which allows customers to pay in online stores with one click while maintaining security.

Pablo Fourez , the firm’s chief digital officer, told The Finance Files that these measures will enhance security and reduce fraud, which builds trust in the brand and allows it to remain competitive in attracting more partners and customers.

Asked whether Mastercard is concerned that the change could lead to more impulsive shopping behavior among consumers, Fourez said that “while simplifying the checkout process could make it easier for people to make more online purchases, our primary goal is to enhance security and convenience for consumers.”

PwC released its 2024 Asset & Wealth Management Report, outlining that 80% of wealth managers say AI will fuel revenue growth by 2028. We talked with Albertha Charles about what the findings indicate for the future of working in asset and wealth management.

Albertha Charles is the global asset & wealth management leader for PwC UK, where she leads and advises PwC’s private equity and asset & wealth management portfolio valuations proposition.

Chang: What do the trends outlined in this report indicate about how asset and wealth management is shifting?

Charles: This industry is looking at a very positive growth environment for the next few years. Our forecast projects that the sector will hit $171.3 trillion by 2028, which is growth of 5.9%, which is higher than last year, when we were looking at 5%. That's very significant. Alternative assets, which have seen huge growth, are actually expected to outpace the market, growing at a higher rate of 6.7%. As interest rates are coming down, search for higher yield is going to keep driving that up. Plus, businesses are staying private for longer, and that's also going to fuel the alternatives space.

When we talk about generative AI and blockchain, we often focus on the productivity gains and the efficiency gains. But this survey tells us that 80% of our asset manager respondents actually expected those technologies to deliver a boost in their top-line revenues. It's really an opportunity. Every CEO and every asset manager is now looking at those technologies and seeing them as an opportunity to think about: how do I redesign my business, how do I access new services, how do I redefine my business model? It's a much broader repositioning and much more transformative in terms of driving growth.

Another key point is that while scale has always been a key driver of mergers and acquisitions in the sector, a lot of the future M&A and partnerships are going to be driven by a need to acquire capability, particularly around tech capability. 73% of asset managers considering M&A see access to skills as the number-one driver of deal making over the next two to three years. That's quite a different shape and dynamic around M&A and consolidation activity in the sector. Traditionally, a lot of it was around driving scale, cutting costs and reducing competition. But this indicates acquiring companies will be more about finding targets with the right capabilities and partnerships to facilitate growth and access to talent.

Chang: How do you foresee finance companies implementing AI technology?

Charles: This is a market where customers are demanding much more customized, transparent, low-cost and hybrid models. They really want a seamless digital experience. And so a lot of asset managers and wealth managers are using those technologies, particularly AI, to analyze customer data, to really understand the profiles, understand the preferences, so that they can tailor really customized services. So, it's really about hyper-personalization, and doing that at scale.

Secondly, they’re offering tech as a service, and there are a number of players who do this already. They develop technologies, and in a number of cases they developed technologies for their own use but they have now changed that into a business model where they are selling this to other asset managers. That's being powered by cloud technology and AI. The way it's helping, if you look at portfolio management, a lot of the new technology there is to predict performance risk of investments so that you have instantaneous, real-time data to help you with rebalancing the portfolio. Firms are doing the same thing with risk analytics. So, you have the ability to analyze a vast amount of data from a diverse array of sources in a way that you've never had before.

Companies are using that to stress-test various scenarios in the market, to really understand what the loss is that you could get on a portfolio within certain market conditions. And they use that to either rebalance the portfolios or inform the hedging strategies. Asset managers who are providing that as a service are offering, if you like, a user subscription model. So, they have really diversified their revenues away from the traditional fee-based models, management fees, which has been under a lot of pressure.

If you look at the leading firms, they have announced divisions within their business model all around offering technology as a service to the market. The larger players are fairly advanced in how they use generative AI and how they leverage big-data analytics. I think the challenge is probably for the smaller players who don’t have the resources to invest in those technologies themselves. With those players, for us it's really about partnership. To remain competitive, medium-sized players are now starting to seek out partners within the ecosystem evolving around those new technologies, largely with fintech and data science companies, to crowdsource the innovation from the ecosystem. That's the way they're going to compete.

Chang: Another thing that the report highlighted is that for asset managers considering M&A, skilled expertise is the number-one driver of deal making, but almost one-third said that many lacked the relevant skills. What does this show about the value of reskilling?

Charles: What is very clear is that asset managers recognize that as technologies redefine the shape of the industry, they need to redefine and reimagine the workforce of the future. That means you've got to reskill, because people are going to have to take on new roles that have been redefined by introducing those technologies. You've got to upskill, but you've got to also look at sourcing different capabilities. We’re really looking at, for instance, data science, data analytics, gen AI, blockchain, and really looking at how you source a different profile to complement the deep asset management expertise that has always been so critical for players in our market. I think there's also a cultural piece to it, and it's really about making sure that people are empowered to experiment with new technologies, and that they are encouraged to disrupt their way of working as a culture. I think that is quite critical in terms of redefining the workforce of the future and getting the full value out of those technologies.

Chang: In what ways do you think these developments will impact the future of jobs and hiring in asset management?

Charles: I think there's a piece about where we source people from and the skill profile that we're sourcing. This is traditionally the sector that has recruited really for expertise as the core differentiator. And this new era is actually about complementing that with the data sciences and with generative AI and blockchain capabilities. Other points that I would highlight are around what our survey respondents regarded as one of the biggest barriers to adopting those technologies and really harnessing the benefits – and that's about people not trusting the outcomes of the technology. It's really quite critical for companies to consider having a very strong cybersecurity framework and governance structure in place, with a particular focus on third-party risk. If you're not able to do your own robust due diligence on your suppliers, it's looking at outsourcing. But being really clear that you are on the framework, and also having a human in the loop, were things that came through in terms of boosting confidence around using some of these technologies. That trust piece goes hand in hand with the people and the workforce of the future.

Here are the latest updates in the world of finance, private equity, banking, real estate, markets and more:

Goldman to spin off crypto platform: Goldman Sachs is holding talks with potential partners as it plans to spin off its digital assets platform, Bloomberg reported. The bank wants to develop the platform’s features and explore new commercial uses, with the goal of launching the spinoff in the next 12 to 18 months. Financial firms have increasingly turned to blockchain technology to manage the issuance, trading and settlement of various assets. Visa unveiled a platform last month to help financial institutions issue stablecoins – cryptocurrencies pegged to the U.S. dollar – to help facilitate business-to-business transactions.

Blackstone buys into Jersey Mike's: The private equity group says it reached a deal to acquire a majority stake in the sandwich chain Jersey Mike’s Subs. Neither company disclosed the terms of the deal, but it would reportedly value the privately owned chain at $8 billion, including debt. Jersey Mike’s is one of America’s fastest-growing restaurant franchises, with about 3,000 locations. The acquisition follows several recent private equity plays in restaurant chains, including Roark Capital’s $9.6 billion purchase of Subway last year.

Bitcoin nears $100K mark: Bitcoin hit a new record above $99,000 on Friday, putting it in striking distance of the $100,000 mark. The cryptocurrency, which broke $98,000 for the first time on Thursday, has set records again and again this month amid crypto holders’ hopes that the incoming Trump administration will provide "more supportive regulation" for the industry. Bitcoin has gained more than 130% this year, and 45% just since this month’s presidential election. It also received a boost from the approval of U.S.-listed bitcoin exchange-traded funds in January, which helped put it "on the cusp of mainstream acceptance," wrote Reuters.

AI experts in banking said that rather than replacing jobs in financial services, AI will be used to drive productivity in the sector. But they say human capital will continue to be a necessity, and finance professionals who have the skills to use AI will see great returns.

We want to hear from you: What impacts will AI have on jobs in financial services in 2025?

Join the conversation in the comments section below.