AI’s Role in Financial Institutions
The rapid evolution of AI is changing the world, and the financial services industry is no exception. Businesses in this sector are striving to accelerate insights, respond faster, predict more accurately, and enhance their overall customer experience.
Of course, the financial services industry remains highly competitive and subject to stringent industry regulations. These industry dynamics strongly influence how technology is adopted within the industry and require financial institutions to continuously identify new ways to differentiate their capabilities using technology.
As a result, there are many compelling use cases for AI in financial services as the industry strives to deliver new value and services around data. Increased use of AI in financial services allows institutions to streamline core business processes while adding innovative products and services that improve customers’ experiences. Financial services companies are also exploring how AI-based enterprise assistants can help their employees be more productive, as well as how AI can be applied to enhance software development.
Most financial services executives expect artificial intelligence to become a pivotal element of success within the next few years. According to a 2021 survey by NTT DATA Services:
- 83 percent [of financial services executives] agree that AI is creating new ways to differentiate offerings and win customers, driven by access to unique data sets.1
- 81 percent said that AI is a critical part of their strategy to attract and retain customers.1
Intel Hardware: Deploying AI in Financial Services
To enable sizable leaps in emerging AI use cases such as fraud detection, document review, risk management, and algorithmic trading, financial institutions of all sizes can rely on Intel hardware, software, and solutions.
With a range of tools to fit needs across the industry, we offer a complete AI enablement portfolio that can help you accelerate results and drive value. Our 4th Gen Intel® Xeon® Scalable processors offer exceptional performance alongside powerful built-in accelerators that are ideal for cost-effectively supporting AI in the financial services industry.
4th Gen Intel® Xeon® Scalable processors offer integrated Intel® Advanced Matrix Extensions (Intel® AMX) to help accelerate deep learning inference and training workloads. In the financial services industry, this technology can be applied to streamline deployment for workloads such as natural language processing (NLP), recommendation systems, and image recognition.
This latest generation of Intel® Xeon® Scalable processors provides a ready-right-now platform for deploying your AI workloads from edge to cloud. And it’s the only x86 data center CPU with built-in AI acceleration.
Additionally, the Intel® Max Series product family—including both CPUs and GPUs—can help unlock advanced data science and AI use cases across the financial services industry. Meanwhile, the AI-specialized Habana® Gaudi® and Gaudi®2 offerings can enable scalable natural language processing with standout performance for deep learning training and inference.
Finally, Intel® Data Center GPUs can be deployed to augment CPUs with powerful parallel processing capabilities to help speed outcomes and accelerate innovation.
Ensuring Confidentiality for Banking AI
Confidentiality, privacy, and compliance are top priorities for financial institutions as they progress on their AI journeys. AI solutions depend on large quantities of data, often harnessed from multiple sources. Organizations in the financial sector need to protect their customers’ data and ensure they align with regulations as they pursue AI innovation, even when they’re sharing information with other vendors or third-party technology providers.
Intel is a leader in hardware-based confidential computing and works alongside partners and customers at the forefront of applying new technologies to help secure financial services AI.
To help secure sensitive data as you enable AI capabilities, we offer a robust toolset of integrated features. These offerings include:
- Intel® Software Guard Extensions (Intel® SGX): Helps protect data in use via unique application isolation technology; selected code and data are protected from modification using hardened enclaves.
- Intel® Trust Domain Extensions (Intel® TDX): Enables deployment of hardware-isolated virtual machines called trust domains (TDs); designed to isolate VMs from the virtual-machine manager/hypervisor and any other non-TD software on the platform.
Our partner ecosystem also plays a critical role in enabling secure AI capabilities in the financial sector. For example, we’ve collaborated with our software partner, Consilient, to create a new, protected approach to federated learning, as well as worked with multicloud security partner Fortanix to create the Intel® Security Solution for Fortanix Confidential AI.
AI in Banking
In the banking industry, artificial intelligence helps companies automate business-critical processes such as risk management and fraud prevention while unlocking new capabilities, such as the use of chatbots and intelligent recommender systems for retail banks. The future of banking will see increased integration and connection between physical and digital platforms—with smarter recommendations for customers and automated detection of fraud and crime.
Fighting Financial Crime
Banks are bound by a complex set of laws and programs that are designed to uncover the financing of criminal activities, both domestically and internationally. For example, the International Monetary Fund, as well as the US and other countries, have established anti-money laundering (AML) regulations, requiring financial institutions to maintain AML programs and report suspicious activities.2
While many of these regulations have proven expensive for banks, the rules have been largely ineffective at preventing or deterring financial crime. Legacy hardware has created a barrier to success, as older systems lack the scale to combat threats and manage complex databases across various business units. Further, AML measures increasingly require real-time analysis to enable faster transactions or support online capabilities. As a result, companies are turning to artificial intelligence to navigate industry regulation and increase efficiency through real-time analysis.
Many banking organizations currently use AI technologies to automate fraud detection at every level across their organization. This is best demonstrated by PayPal, who improved the detection of fraudulent transactions using Intel® technologies integrated into a real-time data platform from Aerospike. Key results included a 30x reduction in the number of missed fraud transactions with a 3x reduction in hardware cost.3
The Connected Branch and Online Platform
Banking organizations are using AI to deliver a holistic customer experience with personalized banking that’s integrated regardless of where customers are—at home, on the go, or in the branch.
In addition to modernizing traditional processes, artificial intelligence can be used to deliver enhanced customer experiences through new services and capabilities. In retail banking, the latest technologies enable banks to understand customers’ needs and offer personalized banking services that are tailored to each individual. Intelligent automation helps streamline the customer experience and speed processes throughout banking organizations. Online chatbots also allow customers to enjoy smoother self-service experiences that can be more convenient than a phone call or in-person visit.
Inside the branch, AI-enabled machine vision solutions help bridge the gap between the physical space and digital channels, including on-site kiosks. For example, machine vision‒based sensors can track customers’ gaze, posture, and gestures; assess wait times; and alert bank employees when a customer needs assistance. These AI-enabled solutions analyze behavioral data from the branch and from online channels. The resulting intelligence is used to individualize and optimize purchasing, placement, and timing of marketing displays and campaigns.4
Across physical and digital operations, AI is also helping banks conduct faster and more-efficient Know Your Customer (KYC) initiatives, which are critical to controlling risks and verifying customer identities. AI-enhanced KYC solutions often include technologies such as biometric identification, intelligent document processing, and real-time transaction monitoring.
AI in Capital Markets
Artificial intelligence is also being used by financial institutions operating in capital markets—asset managers and hedge funds, among others—to improve efficiency and deploy new capabilities. AI technology is often used to support risk management processes in addition to optimizing trading strategies for a variety of financial instruments.
Liquidity and Risk Management in Trading
AI can help investment banks and other financial institutions comply with a new set of international regulations called Fundamental Review of the Trading Book (FRTB). Beginning in January 2023, financial institutions will need to calculate all risks associated with their trading positions in securities, commodities, foreign currencies, and other investments. Because of the massive scale, FRTB compliance will rely on complex financial modeling, simulations, and impact studies that require enormous investments in computational power and data storage capacity.
Artificial intelligence can be used to significantly increase the speed at which this analysis is completed. For example, software vendors such as Matlogica and Quantifi achieved significant performance improvements through a variety of valuation adjustment (xVA) models based on machine learning and deep neural networks. These AI-enabled enhancements help capital markets companies remain compliant while significantly improving the efficiency of their risk models.
Within capital markets, AI is also enabling new capabilities, including real-time analysis that supports algorithmic trading. Financial trading is based on patterns that are revealed in a history of market behavior and transactions. Recently, companies have begun using AI capabilities to deploy algorithmic trading that relies on machine learning, neural networks, and predictive analytics to interpret and respond to market signals within microseconds.
According to a 2020 JPMorgan study, over 60 percent of trades over USD 10 million were executed using algorithms. The algorithmic trading market is expected to grow by USD 4 billion by 2024, bringing the total volume to USD 19 billion.5"
While algorithmic trading is not new, today’s AI capability accelerates the near-real-time analysis needed for traders to remain competitive.
AI in Insurance and Payments
Finally, artificial intelligence is being used by insurance and payment companies to automate processes, improve efficiency, and deploy new capabilities.
Underwriting and Claims Management
Within the insurance industry, companies are deploying predictive models to streamline the underwriting and claims management process with artificial intelligence.
During customer onboarding, insurers can assess an applicant’s risk factors at a given time. These increasingly sophisticated models rely on machine learning to analyze a variety of factors (e.g., credit, health) to offer a customized premium for their insurance services. Once a customer is onboarded, insurance companies are using AI to receive and process insurance claims with high performance and accuracy. This enables customers to receive insurance services quickly and efficiently. These processes are enabled by robotic process automation technology, which is a machine learning technique that enables hyper automation of various tasks.
Like their counterparts in banking, insurance and payment companies are deploying fraud detection based on natural language processing algorithms to automatically help detect criminal activities—or even predict them before they happen.
Payment processors and credit card issuers also deploy recommendation engines to predict the preferences of customers and prospects. The institutions then offer personalized banking services to those prospects whose demographic profile and behavior either follow a discernable pattern of their own or resemble a similar group whose behaviors are known.
The machine learning‒based recommendation engine analyzes vast amounts of preference data to choose the best fit between product and prospect. These engines are similar to those used in e-commerce stores or streaming media services that recommend additional items based on an individual’s past purchases and on related purchases by other customers with a similar history.
Intel Enterprise Software Products
While Intel is commonly known for its hardware and processing offerings, we also offer powerful software tools that help enable the seamless adoption of artificial intelligence.
To help manage the cloud environments that are critical to many financial services AI solutions, we offer a portfolio of tools that can increase the performance and efficiency of cloud resources—including Granulate™ by Intel and Intel® Cloud Optimizer by Densify. And to help financial services organizations meet their sustainability objectives, our integrated Intel® Data Center Manager can be used to supply real-time data center information to AI energy optimization and insights solutions.
We also offer convrg.io, a full-stack machine learning operating system that’s ideal for multicloud AI and machine learning development. This powerful platform enables AI developers to: manage data and model versions, build complex flows, execute them on multiple machines or clusters, monitor the progress, compare the results, execute as a service, and build an application.
To continue the advancement of confidential computing, we’re working on Project Amber, which offers a new, innovative approach to third-party attestation. It’s a SaaS-based tool that provides remote verification of the trustworthiness of a compute asset based on attestation and policy. Initially, Project Amber will verify the trustworthiness of Intel trusted execution environments (TEEs), but the vision extends to much broader device verification, like IPUs, GPUs, platform roots of trust, and beyond.
Intel® technology is also optimized with the largest cloud providers and hundreds of commercial software vendors, and we continue to participate actively in the open source community, including the Linux Foundation and FINOS. These efforts have resulted in a broad array of partner solutions that help financial institutions accelerate their AI performance and improve their time to business value.
Intel AI Developer Resources
Intel also offers a number of AI developer resources and purpose-built optimizations that can help simplify development, streamline deployment, and maximize performance.
The 4th Gen Intel® Xeon® Scalable processor is optimized for the most popular data science tools and libraries, enabling practitioners to build and deploy their own AI solutions. Our optimizations with PyTorch, BigDL, TensorFlow, and scikit-learn, as well as resources such as the Intel® Distribution of OpenVINO™ toolkit, enable developers to scale their AI environments seamlessly across nodes from edge to cloud while delivering standout performance.
Plus, the Intel® Developer Cloud platform provides easy access to cloud-based hardware resources that can be used to test financial services AI solutions. This tool gives developers an easy platform to learn, prototype, and test as they develop new AI innovations for the financial service sector.
Your Strategic Partner for Financial Services AI
Intel has been working with financial services companies for decades to help them address their most complex AI and analytics challenges.
As a leading technology innovator, we serve as a trusted partner to financial services institutions that are interested in deploying artificial intelligence within their organizations. This experience is critical to ensuring that the financial services industry has the tools and resources it needs to compete globally.