Financial Companies’ Strategies to Acquire AI Capabilities

    Insights » Global » Financial Companies’ Strategies to Acquire AI Capabilities

    Although they are already experimenting with new AI capabilities, banks and financial companies are still exploring how to build the ideal tech structure and governance for them. Spoiler: there is no single path to this.

    AI in banking and finance
    Fabiane Ziolla Menezes

    Fabiane Ziolla Menezes

    Content Strategist

    Contact the author

    The rise of generative artificial intelligence is offering organizations the opportunity to reinvent their businesses, with financial companies experimenting it to enhance front-end and processing capacities. The emergence of Large Language Models (LLMs), especially OpenAI’s ChatGPT in 2022, helped turn this family of technologies, one subset of GenAI, into an everyday utility for several sectors. Even those already working with machine learning and other forms of traditional AI tools see new opportunities, especially when it comes to improving customer service through a whole new level of customization. Still, they have a long way to go before achieving their ultimate (or promised) potential.

    As a recent paper1 from the Organization for Economic Co-operation and Development (OECD) pointed out, “despite the hype around GenAI, advanced use cases of AI in financial markets involving full end-to-end automation without any human intervention remain largely at development phase, if any.”

    Even global players such as Visa, with a 30-year experience in using AI, are in a moment of experimentation

    Even global players such as Visa, with a 30-year experience in using AI, are in a moment of experimentation. The company is currently studying how to advance the predictive capabilities of its over 150 proprietary models. In a recent report, it said2 it “has been deploying cross-functional teams,” comprised of “several thousand” employees “to review how best to deploy generative AI to drive productivity, develop new products and services.” It has, for example, enabled “a secure instance of GPT-4” to its teams.

    While experimenting, banks and financial firms are struggling to find their way around their governance and infrastructure needs surrounding AI tools, which include determining cloud processing architecture (private, public or hybrid) and decision-making processes (centralized or decentralized).

    FIs can approach this in different ways, but according to McKinsey, they should be looking first at building a hybrid cloud setup3 (a mix of private and public cloud servers and on-premises data centers). It’s not that new AI capabilities can’t begin to be developed on private servers, but they can only gain scale or flexibility within a cloud structure. Furthermore, it would be a waste not to do so in such an environment since new generative AI tools could reduce migration time to the cloud by around 30-40%4.

    AI tools could reduce migration time to the cloud by around 30-40%

    However, this is an ongoing process. A survey5 of 100 global banks by Accenture, a leader in digital transformation working as a partner of the main cloud computing services, indicates that only 15% of global banking workloads were running in the cloud in 2022, up 9 percentage points from a year earlier. Most of the workload migrated is in functional areas such as enterprise (35%), data & analytics (20%), and surrounds (21%), which is how Accenture calls the technology for customer interactions such as ATMs, online and mobile banking, and call centers. On the other hand, core activities such as payments and risk and compliance do not surpass 12% of the workload migrated. Among the institutions surveyed, 34% are in North America, 31% in Europe, and 35% in emerging markets such as Brazil and India.

    It’s no wonder banks’ spending on cloud computing tends to grow 15% a year through 20266, outpacing their overall investments in IT, according to market researcher IDC.

    Most FIs, according to McKinsey, given their regulatory burden, are experimenting with AI in a more centralized way7. The OECD paper also states that FIs tend to deploy versions of foundation models “offline,” that is, “within their firewalls, or at the private cloud of their firm, with a view to ensure both data sovereignty and security and comply with existing frameworks for privacy, security and model governance.” Research and development decisions, as well as risk management related to AI also tend to be more centralized.

    What Financial Institutions’ Modern Tech Stack Looks Like

    Before introducing how FIs and financial companies are using AI and GenAI, we must first differentiate between the two technologies. As explained above, GenAI is a subset of AI. But while previous AI models had been trained with a limited dataset and for specific purposes such as pattern identification, classification, and prediction, something elementary in finance, GenAI models can create something new or significantly modify content (text, images, videos, audio, or software code) by aggregating unstructured data. It does that in response to prompts in ways that very much mimic human-generated work. In other words, while old AI tools analyze data and make predictions based on a pre-defined strategy, GenAI invents new data similar to its training set.

    The modern banking technology stack is in full development, but some common features have already been identified as must-haves, many of them inspired by born-digital challengers, as traditional financial institutions mirror them to find new ways to revamp their legacy systems and build new solutions. As suggested by the OECD, AI current usage can be divided into two organizational layers:

    Back-office/Middle-office. Most AI and LLM solutions in use are related to the automation of processes, which are aimed at enabling efficiencies and improving productivity at the operational level. Profit and loss (P&L) procedures and other reconciliations can be replaced with faster and cheaper automated ones. GenAI applications are also being deployed to enhance compliance and risk management tasks, assisting in summarizing documents and norms, for example, and human resources processes.

    Front-office. In addition to operations, AI-driven tools are expected to disrupt sales and marketing, as well as customer support, being used for individualized communication with clients and suppliers. It is also likely to help develop new customer-centric products and services, as it can accelerate the identification of pain points and their respective solutions.

    AI Use in Finance

    When analyzing the digital front of FIs, the latest report8 on banking technology from the Brazilian banking association Febraban in partnership with Deloitte, gives an idea of how AI and GenAI are being used in the software development stages:

    How Brazilian FIs are using AI and GenAI

    Coding, Analytics, and Hyper-personalization as Imminent Use Cases

    GenAI tools can support and accelerate software development, especially as an assistant to developers. This use is pointed out as a top priority by enterprise executives from different industries interviewed by KPMG9 in 2023.

    Top priority functions for GenAI adoption

    Regarding finance, 36% of the 22 biggest FIs in Brazil are using AI for software coding development. It is in this kind of use, however, that many of the main risks linked to GenAI appear, including biases and the so-called ‘hallucinations,’ which are misinterpretations of data arising from various factors, such as training data preconceptions or inaccuracy. Although heavily investing in the area, it may take some time for the finance sector to solve the dilemmas involved in using AI for coding.

    New GenAI tools are also being used to unfold existing analytics applications into new ones, as they can work with unstructured data, extracting10 from them things “like historical service interactions, social posts, news, and web pages and provide frontline bank employees with prompts that enhance their engagement with customers.” As pointed out by OECD’s report, this could be transformative in fields like credit analysis. 

    On the customer service front, prior AI solutions were primarily used to power chatbots and automated call centers. GenAI is now making these interactions more similar to human-like conversations. This is just part of a hyper-personalization trend that can also include customer segmentation at the individual level, from application interfaces that showcase products and services that best fit customer profiles to personalized investment recommendations, but in a much faster (almost in real-time) way.

    FI Investment in GenAI

    A new study from Juniper Research indicates11 that investments in GenAI by banks will reach USD$ 85 billion in 2030, from USD$ 6 billion globally in 2024. It is not yet clear, however, how much will be spent on buying third-party ready-to-use solutions and how much will be spent on in-house research and development. According to S&P Global, this choice depends on their scale and investment capacity12.

    From an investment point of view, it would make much more sense to partner with a key GenAI player to speed things up, rather than trying to build solutions from scratch. S&P Global lists some examples in this sense, such as Morgan Stanley’s wealth management division partnership with OpenAI’s GPT-4. The bank is using the technology to help its employees locate relevant in-house intellectual information and cross it with company insights across sectors and regions, and information on asset classes. Wells Fargo, in turn, has partnered with Google. It is using Dialogflow, Google’s conversational AI, to empower its virtual assistant Fargo.

    Major cloud computing vendors developing GenAI processing and applications, Amazon Web Services, Google Cloud, Microsoft Azure, and IBM are also central players in the space, as they provide AI-focused services, including LLMs, to boost solutions development in this field. Just to stay within the examples already given, Morgan Stanley named13 Microsoft its top cloud partner three years ago, and Wells Fargo, in addition to Google, is also migrating software applications to Microsoft Azure14.

    Along the way, more partnerships between financial companies and big tech will emerge, which should favor, to some extent, the faster development of FIs from more advanced markets. This does not mean, however, that emerging market financial institutions will be left behind. The largest publicly listed bank in Latin America, Itaú, has more than 300 employees dedicated to finding ways to use AI to improve current services.

    Next Steps

    This new wave of AI-driven solutions promises to reshape the financial industry increasingly and incrementally. Several use cases are emerging, and we plan to explore some of them in a second article of this special AI series. If you have some specific inquiries about AI and GenAI in banking or payments, we can help you get answers. Contact us to learn more!


    Keep up to date with our e-commerce, payments and crypto insights:


    Sources

    Fabiane Ziolla Menezes
    Fabiane Ziolla Menezes
    fabiane@paymentscmi.com

    Fabiane is a journalist with more than 15 years of experience reporting on business, finance, innovation, and cities in Brazil. At PCMI, she keeps an eye on revolutionary fintech movements and the businesses behind these big changes in Latin America.

    Post Tags: