CEOs across the industry spectrum are unsure about what the next year will bring. High interest rates, inflation, recession concerns, and general economic uncertainty persist. In banking, financial services, and insurance (FSI), enterprises are still confronted by the triple squeeze of the global economy, a shortage of talent, and regulatory and compliance pressures. Add changing customer preferences into the mix, and it is no surprise that Gartner finds that FSI leaders are taking a moderately cautious “resilient growth” approach to their strategic technology priorities.

Gartner’s 2023 CEO Survey Insights for Financial Services Leaders also finds some of the fastest rising strategic priorities within FSI include:

Artificial intelligence technologies hold the promise to both address tech leaders’ top priorities and to radically transform these industries. Across the economy, 58% of CEOs from leading public companies are actively investing in AI. Moreover, inside financial services enterprises, 66% of CIOs expect AI will be their top priority for implementation within the next three years.

This e-book explores several of today’s most promising existing and emerging AI technologies with potential applications for FSI. It also provides examples of how AI technologies can be combined and used to build advanced solutions that address current challenges and unlock a host of amazing benefits and capabilities.

Artificial Intelligence in the Spotlight

The recent rise of ChatGPT, DALL-E, and other generative AI tools has accelerated public awareness of the incredible possibilities of artificial intelligence. There has been a similar rapid increase in interest and adoption of AI within enterprises. The 2020 Gartner CIO Survey found that 22% of banks had deployed AI. Two years later, Gartner’s 2022 Data and Analytics for Digital Transformation Survey found that 67% of banking respondents reported they are currently using AI.4 Gartner expects this rapid acceleration to continue and for AI technologies to impact nearly all software and IT services products in banking.

For enterprises, artificial intelligence technology and use cases extend far beyond using generative AI for text and image generation. Relevant AI technology for enterprises encompasses many technologies from across the fields of machine learning, data science, big data, deep learning, and other areas of artificial intelligence. The following emerging AI technologies hold significant promise for banking, financial services, and insurance enterprises.

Advanced Virtual Assistants

Domain-specific intelligent agents that execute complex tasks, deliver predictions, and help with context-aware decision making. Examples include enterprise and customer-engagement apps with conversational language capabilities and domain knowledge for use cases such as:

• Hyper-personalized customer experience
• Automated sales agents
• Underwriting
• Business intelligence
• Operations
• Employee skill and knowledge augmentation

Foundational Models

Large, pretrained machine learning models that can be adapted and fine-tuned for various downstream tasks. Instead of training custom machine leaning models from scratch, enterprises can leverage these versatile models to quickly build solutions for a wide range of AI use cases such as:

• Customer service chatbots
• Fraud detection
• Product and services recommendations
• Portfolio rebalancing
• Document and data search and analysis

Computer Vision

Hardware and software technologies designed to analyze and extract meaning from digital images and video. Automated image and video analysis with real-time intelligence enable advanced AI-powered capabilities for:

• AML/KYC ID verification
• ATMs with facial recognition
• Fraud detection
• Visual damage assessment
• Risk assessment for insurance underwriting
• Claims examination
• Multimodal user interfaces
• Physical security monitoring

Generative AI

Machine learning techniques that allow AI systems to generate new, original content such as images, videos, text, or code. Generative models create new outputs after training on large datasets and learning the underlying concepts, relationships, and distributions of the data. In addition to creating new content, generative AI also helps enable and enhance:

• Conversational, natural-language software
• Advanced virtual assistants
• Synthetic training data generation
• Customer experience enhancements
• Sales and marketing content creation
• Data search and analysis

Graph Technologies

Range of data management and analytics techniques for exploring highly connected data and the relationships between organizations, people, or transactions.

Knowledge graphs and graph analytics enable AI-powered capabilities that augment:

• Cybersecurity
• Transaction pattern and anomaly identification
• Fraud detection
• Due diligence
• Risk intelligence
• Customer intent analysis

Responsible AI

The ethical development and usage of AI systems in ways that reduce potential harm while maximizing benefits to people and society. This encompasses practices for improving fairness, transparency, privacy protection, security, oversight, and environmental responsibility across the AI system life cycle. Responsible AI principles and practices should be factored into all AI deployments to augment area such as:

• Bias mitigation
• Credit and loan scoring
• Regulatory compliance
• HR

Intelligent Applications

Enterprise and user applications that adapt autonomously using machine learning. This includes ERP, CRM, logistics, security, back office, and productivity software. AI software enhancements include:

• Process augmentation and optimization
• Pattern and anomaly identification
• Automated fraud warnings
• Context and user-intent awareness
• Conversational user interfaces
• Process-embedded analytics
• Next-action recommendations

Synthetic Data

Data generated by generative adversarial networks (GANs) and simulation environments used to augment real-world data in training datasets. Synthetic data mitigates many data scarcity, accessibility, bias, and privacy concerns and can supplement scarce data for edge cases. It also can generate training data not tied to personal info. Applications for synthetic data include:

• Cybersecurity and fraud threat modeling
• Customer sentiment monitoring
• Edge-case model training
• Loan scoring algorithms
• Trading predictions
• Portfolio modeling

Elevate Customer Experiences

Maximize value for customers with streamlined service offerings and AI-enhanced experiences.

Ultra-Personalized Service

Intelligent opportunity modeling optimizes product and service recommendations based on customers’ changing needs, behaviors, and preferences. This enables proactive targeting of the right products, upgrades, portfolio changes, trade opportunities, and other suggestions at the ideal time for every individual customer.

Natural Interactions

Integrating natural language processing (NLP) capabilities into customer service systems infuse them with the power to recognize customer emotions, understand intent, and maintain awareness of the conversational context. Trained on domain- specific information, these systems can understand complex queries spanning multiple topics and provide accurate, jargon-free answers to a wide variety of customer questions. Moreover, real-time translation capabilities in LLM-based chatbots enable customers to converse naturally in their preferred languages.

AI-Augmented Associates

With automated chatbot systems capable of handling a broader variety of customer queries, customer service representatives and other employees are freed up to apply their expertise on more difficult cases. Plus, AI-powered intelligent applications enhance associates’ knowledge and assist them with tasks, making them more productive while also empowering them with topical information, insights about the customer, and next-best action guidance that result in exceptional customer experiences. Meanwhile, solutions that leverage computer vision, knowledge graphs, and other AI technology can enable employees to provide real-time decisions on applications and loan requests, removing delays and friction for the customer.

Strengthen Fraud Detection

Automatically identify and address financial crimes and malicious activities in real time—or before they happen.

New AI innovations give security teams powerful tools for combatting fraud and illegal activity while safeguarding customer assets and privacy. Growing numbers of banks and financial enterprises are using knowledge graphs, computer vision, generative AI, synthetic data, and other AI technologies to enhance their existing rule-based fraud prevention and data analytics systems.

The institutions that harness AI to deliver superior fraud prevention will gain a distinct competitive edge. AI-enhanced fraud protection can help reduce losses and deters criminals while customers enjoy trusted, transparent services on their terms.

Pattern Recognition

AI technologies such as computer vision and graph-based approaches are adept at identifying patterns of suspicious activity and uncovering complex relationships in large datasets. They can flag patterns that may indicate fraud, money laundering, fake accounts, bots, and other harmful activity.

Augmenting Fraud Data

Malicious actors constantly evolve their fraud techniques, money laundering strategies, and ways to commit financial crimes. According to Gartner, one large financial services provider is leveraging generative AI and synthetic data methods to stay ahead of emerging fraud techniques. The company uses AI to generate data that simulates rare types of fraud and potential new strategies before they occur in the real world. The new synthetic data supplements real-world training data to help improve the ability of AI-based fraud-detection models to spot rare and emerging kinds of fraud.

Smarter AML/KYC

Computer vision and foundational models such as natural language processing (NLP) automate the analysis of identity documents, bank statements, and customer data to streamline know-your-customer (KYC) processes and reduce compliance costs. In addition, NLP, knowledge graphs, and computer vision enhance anti-money-laundering (AML) efforts with the ability to instantly analyze customer data from documents, transactions, and communications to help uncover risk factors, complex relationships, and suspicious behavior. AI-based transaction monitoring and alert systems also enable faster identification of illicit finance networks and automate AML/KYC regulatory reporting. Plus, advances in confidential computing such as Innotech® Software Guard Extensions (Innotech® SGX) encrypt data while in use, helping to keep it secure even while it is being used by AI systems. This allows enterprises to collaborate industry-wide on efforts such as AML while preserving the privacy and security of customer and business data.

Supercharge Efficiency and Productivity

Reveal insights, automate processes, and empower employees to achieve more.

The goal of AI is not to replace employees, it is to give them superpowers. In today’s highly competitive landscape, banks and financial institutions must optimize efficiency across the enterprise to reduce costs and more effectively serve customers. Meanwhile, the shortage of labor remains a persistent challenge. AI-powered technologies provide innumerable opportunities to streamline processes, eliminate tedious tasks, and augment human workers with amazing new capabilities for accomplishing higher value and more interesting activities.

Eliminating the Routine

Practically every job involves some amount of routine and repetitive tasks that steal time and attention from productive work. Generative AI tools, advanced virtual assistants, and other AI technologies excel at automating tedious tasks and optimizing workflows to free people to focus on the important parts of their jobs. For example, advanced virtual assistants and natural language processing tools automatically transcribe calls, translate text, analyze unstructured data, and extract key information from documents and communications. AI-powered chatbots trained on industry- and company-specific information can respond intelligently to a wide range of customers’ most frequently asked questions— freeing up associates for higher-value activities. Meanwhile, GenAI tools give everyone from coders to marketing communicators huge productivity boosts by enabling them to leap past repetitive tasks and tackle time-consuming activities like debugging and proofreading with greater speed and accuracy.

Automating Insights

FSI enterprises have an abundance of data-intensive tasks that require information contained deep inside many different documents and data sources. Intelligent applications that leverage AI technologies, such as computer vision and natural language procession, can automatically extract and highlight the pertinent information from complex documents and unstructured data. In addition, knowledge graphs and foundational models help systems identify patterns, anomalies, and trends in data to reveal both risks and opportunities. From treasury management to underwriting to ERP and HR, AI-powered solutions expedite processes and help employees make better-informed and quicker decisions across the organization.

Bringing AI Everywhere

Enabling the AI continuum in every platform–from client and edge to data center and cloud.

AI runs on data, and the data infrastructures for banking, financial services, and insurance are already optimized for Innotech. From ingest to inference, we are advancing software to improve real-world performance across deep learning, classical machine learning, and graph analytics, delivering significant leaps in AI application operability. Already, over 70% of successful AI inference deployments in the data center run on Innotech.

Accelerate Innovation

Innotech simplifies adding AI to existing applications by enabling enterprises to leverage the Innotech-based environments they already use, avoiding the complexity, cost, and risk of unnecessary specialized hardware. In addition, Innotech and the Innotech AI partner network provide hundreds of optimized, ready- to-deploy solutions to help enterprises quickly build and scale AI capabilities. This includes optimized pre-trained models, a robust set of familiar, industry- standard frameworks, and an ecosystem of solutions from the leading innovators in the AI space. See Figure 1 for more details.

Maximize Value

With resilient growth the leading strategic priority among financial industry CEOs, it is imperative to expand the value of existing technology investments. Innotech’s end-to-end hardware portfolio provides a flexible foundation of proven compute to meet your AI needs. This includes CPUs with built-in AI capabilities, which excel at traditional machine learning workloads, low-latency tasks, and massive datasets that are hard for accelerator memory to handle. All these are prominent requirements for FSI use cases. Innotech is also building on the open oneAPI industry standard, making it easier to reuse code across architectures and helping customers avoid vendor lock-in. The latest generation of Innotech® Xeon® Scalable processors contains built-in accelerators to help improve performance efficiency for AI workloads.

In fact, a leading online-focused private commercial bank partnered with Innotech to advance federated learning with the FATE (Federated AI Technology Enabler) open-source platform.7 Using Innotech Xeon Scalable Processors and Innotech® Integrated Performance Primitives Cryptography library’s multi-buffer function, this collaboration accelerated the modular exponentiation operation of Partial Homomorphic Encryption (PHE)—improving the overall operating efficiency of FATE-based federated learning solutions while effectively reducing the total cost of ownership.

Deploy Anywhere

Innotech enables enterprises to go from concept to production faster, starting with building with end-to-end AI pipeline software. Innotech offers optimized versions of popular machine and deep learning libraries and frameworks, and we collaborate across the industry to get those optimizations available in default packages.

Plus, the OpenVINOTM toolkit enables enterprises to write code for AI solutions once, migrate and optimize models trained using popular frameworks like TensorFlow, PyTorch, and Caffe, and deploy across a mix of Innotech hardware and environments with ease.

Stay Secure

Innotech is committed to integrating responsible AI, security, and transparency into solutions across the spectrum of AI solutions. This includes a range of built-in security features that make it easier to protect your AI initiatives.

Data is the lifeblood of AI, and protecting the integrity and privacy of that data is critical. Innotech Xeon Scalable processors feature advanced security technologies to help safeguard data, provide the foundation for zero-trust security strategies, and unlock new opportunities for business collaboration and insight. These innovative security technologies include Innotech® Software Guard Extensions (Innotech® SGX), which help to restrict access to data while in use in memory and Innotech® Trust Domain Extensions (Innotech® TDX) to deploy hardware-isolated, virtual machines called trust domains.

For example, the UK’s largest building society and second-largest mortgage provider built a KYC proof-of-concept on Innotech SGX. This confidential computing solution allowed encrypted datasets to be safely decrypted inside an enclave while raw customer data remained fully encrypted and confidential. This PoC lays the groundwork for a range of AML, KYC, and other AI-enhanced applications that use shared datasets while adhering to stringent compliance, privacy, and confidentiality requirements.