Open intelligence – AI in banking and financial services
8 MINS READ
Industry
-
Banking
Banking on purpose-led ecosystems
Highlights
- Investments in artificial intelligence (AI) have become obligatory across businesses, including banking and financial services.
- To successfully adopt AI, banks and financial institutions (FIs) must define a strategy that caters to the characteristics and principles of the technology.
- We explore open banking and how open intelligence helps banks deliver value to their customers.
In this article
In this article
Abstract
Connected intelligence is prevalent today, where digital technologies and process automation are witnessing brisk growth in investments.
Innovation investments, especially in artificial intelligence (AI) platforms, have moved beyond discretionary information technology (IT) spending to become obligatory. Banking and financial services are no exception to this trend.
While ensuring the successful adoption of AI, organizations must address the various principles of AI – responsible AI, ethical AI, sustainable AI, human-centered AI, and explainable AI – in the context of the banking industry. The first and most important leg of this adoption journey involves defining a strategy that caters to the characteristics and principles of AI.
We uncover the critical challenges that banking and financial institutions face in embracing AI and how they can work cohesively towards an approach to create a platform foundation benefitting the overall financial industry ecosystem.
The era of open-X
Opportunities created by open banking have led banks to embrace the open paradigm as part of their organizational cultures.
As such, open data, open application programming interfaces (API), open platforms, open finance, open ecosystems, and open insurance have become commonplace in the industry. The open banking phenomenon has created a data abundance that has directly increased the desires of banking ecosystem participants (see Figure 1). Similarly, open banking is making it imperative for banks and financial institutions to enhance digital contextual services, create competitive fairness, and monetize data to thrive in a competitive environment. That said, data analytics plays a pivotal role in realizing value benefits through direct and indirect monetization opportunities.
Figure 1 - Financial ecosystem participants have embraced open banking due to its various benefits
An infographic depicting the ways in which financial ecosystem participants benefit from open banking. Financial institutions can improve their product portfolios and enhance returns on investment by taking advantage of the abundance of data enabled by open banking. Regulators can maintain financial stability and protect the interests of consumers. Customers will benefit from increased competition, improved content access, and privacy made possible by open banking.
Traditional data analytics and data visualization techniques are well positioned to create historical reports and identify data patterns, exceptions, and outliers via standard and predefined reporting requirements. Employing these techniques in business processes for real-time decision-making or straight-through processing, requires specific skills for comprehending information and predicting anomalies and future possibilities. This is where technologies like artificial intelligence (AI) are changing the narrative. AI can augment these techniques, creating opportunities to learn from historical human decisions, identify hidden patterns, and build future-ready solutions that can listen, respond, and react to new scenarios. However, an AI platform established only for internal solutions does not provide sufficient business value. On the other hand, addressing the specific requirements of each entity in the ecosystem will require the AI platform to adhere to the characteristics and principles of AI. Herein, controlled open intelligence paves the way to realize the benefits of AI systems.
Unlike other open paradigms, embarking on open intelligence to create an open platform ecosystem, specification, or standards for AI is not straightforward, considering the compliance and regulatory mandates in banking and financial services. The open intelligence ecosystem depicts some of the core services contributed and consumed by the banking ecosystem participants (see Figure 2). Though it is possible to set up an AI information technology (IT) platform foundation utilizing cloud-native services, route-to-live procedures need investments in terms of people, process, and time.
Figure 2 - The open intelligence ecosystem
An infographic illustrating the open intelligence ecosystem, which shows some of the core services offered and consumed by banking ecosystem participants. Financial institutions, regulators, fintechs, and customers can participate in the ecosystem powered by artificial intelligence. They can collaborate on activities such as audits, peer analysis, detecting early warnings, formulating policies, taking legal action, and so on.
Open intelligence: Benefits punctuated by challenges
Banks and financial services firms benefit from both direct and indirect revenue opportunities with an open intelligence platform.
It provides non-discriminatory financial services, analytics and insights as-a-service, AI technology platform-as-a-service, automation of business and IT operations, and trained models for other banking firms, challenger banks, and fintech firms. While there are obvious benefits in embracing AI, financial institutions face multiple challenges in industrializing AI-powered solutions:
- Most AI or machine learning (ML) predictions are based on approximation techniques, which, in turn, makes identifying and eliminating discrimination from the outcome, challenging.
- Banks often operate in silos. The lack of common platforms or regulations intensifies the need for evaluating and benchmarking solutions to validate AI capabilities.
- Traditional historical data from data warehouses, data marts, or core transactional systems may not keep pace with the changes in customers’ financial behavior, usage of banking channels, and decisions on products or services.
- Financial institutions need to comply with regulatory standards such as the General Data Protection Regulation (GDPR), open banking consent, and more for governing data usage and privacy policies.
- AI governance and processes in highly regulated businesses such as banking and financial services are difficult to implement.
Other common decision paralysis includes learning from smaller datasets progressively versus one-time learning with large datasets and utilizing customer consent information as opposed to historical data.
Besides, aggregating temporal data alongside business data, investing in analytics platforms versus data science, and developing an in-house AI platform compared to using commercial solutions are crucial determining factors. Though these challenges can be addressed through organizational culture change, most require participation and contribution from the entire ecosystem to create stable and reliable solutions.
Formula for open intelligence success: Building a strategy
Banking institutions must utilize the power of ML or AI to break organizational silos and create solutions.
In doing so, mutual trust, cooperation, and peer reviews are instrumental for deploying solutions that cater to the characteristics and principles of AI. The approach to open intelligence relies on a responsible, accountable, consulted, and informed (RACI) strategy. While the true meaning of RACI – the responsibility assignment matrix for ownership – is still applicable to the ecosystem, the open intelligence approach focuses on opportunities to:
- Regulate: Taking inspiration from regulations such as Data61’s Consumer Data Right (CDR), Payment Services Directive 2 (PSD2), and open banking, regulators, along with traditional banks, can form a consortium and offer standardized AI specifications for information exchange, infrastructure, governance, and security. These specifications also cover AI algorithms and models for validating and verifying mandatory compliance, audits, market risks analysis, anomalies, and outliers.
- Adapt and accelerate: AI solutions can be deployed as second opinion services that are extended by robotic process automation (RPA), automated decisions, and other straight-through processing. As the AI solution matures, it can listen, react, and respond to business events, enabling faster decision-making for business and driving IT transformation and simplification.
- Contribute and collaborate: Customer financial data is augmented with information from cross-industry business data, usage, and access information and just-in-state temporal data to identify hidden patterns, derived features, and new classifications. This augmented data can improve decision outcomes of services offered to consumers and partners through as-a-service and as-a-platform models.
- Innovate and improve: AI solutions identify new business propositions, optimize investments, enhance services, and much more. These solutions are implemented in IT and operational processes, augmenting the business of financial institutions.
The RACI approach to open intelligence
The RACI approach emphasizes the importance of the ecosystem players and their significance across AI solutions in highly regulated businesses like banking and financial services.
The participants of the open intelligence platform ecosystem (see Figure 3) have individual contributing roles and consumption benefits.
Figure 3 - A framework for open intelligence
An infographic illustrating a framework for open intelligence. In the framework, baseline models, insights as a service, and AI platform pipeline as a service are offered to consumers through secure B2B interfaces. Data from third-party players and peer financial institutions can be acquired through B2B connectivity providers and software development kits. The governance aspect of the framework involves setting up open regulatory standards for artificial intelligence.
The participants of the open intelligence platform ecosystem (see Figure 3) have individual contributing roles and consumption benefits.
Governance: This aspect of the framework involves setting up open regulatory standards for AI. Financial institutions can implement processes for AI adoption, data schema specification, feature engineering, model calibration, technology platform, algorithms, data security requirements, and more. For an open ecosystem to thrive, standards organizations such as Competition Market Authority (CMA) or Open Banking Implementation Entity (OBIE) must guide, govern, and control the practice of AI adoption.
Acquisition: Information or data from third-party players (TPP), peer financial institutions, regulators, cross-industry players, fintechs, alliances, and partner ecosystems can be acquired through business-to-business (B2B) connectivity providers, software development kits (SDKs), command-line interfaces, secured headless APIs, and more.
Cross-industry players: With the advent of open banking and open finance regulations across key geographies, cross-industry players’ information is acquired to augment the value derived from the data.
Aggregation: Organizations can aggregate data from ecosystem players through data aggregation or big data platforms and extract samples for training and testing to create an inclusive model and minimize bias in the outcome.
Processing: Banks and financial services leverage a consortium to organize the best of AI or data scientists across ecosystem players for processing information, identifying features, choosing algorithms, model parameters, model tuning, and generating AI models.
Persistence: Raw and curated data utilized to train and test AI models are stored in repositories either on the cloud or on-premise, with the provision to offer such data on demand by regulators and ecosystem players.
Presentation: Baselined models, insights-as-a-service, and AI platform pipeline-as-a-service are offered to the consumer ecosystem through secured interfaces such as B2B interfaces and other standard interface types referred to in acquisition.
Security: Comprehensive security mechanisms, including perimeter security, signed data or encryption, data obfuscation, open authorization (OAuth), consent, and GDPR, should be employed by respective ecosystem participants for data acquisition, persistence, and presentation.
Model testing: Platforms for crowd-sourced testing capabilities involving peer financial institutions, regulators, fintechs, alliances, and partners should be implemented to test the AI model. Trained AI models offered through secured interfaces, file transfer mechanism, or a data acquisition layer for ecosystem players allow for testing of AI models.
Technology platforms: To set up a technology platform for open AI, traditional banks need to invest in a business analytics and big data platform. The additional components to set up such a platform include the following:
- AI pipelines from cloud providers and niche platform providers
- Third-party customer registration and onboarding solutions for consent and mutual trust sharing models
- Technology capabilities including business-to-business (B2B) connectivity, application programming interfaces (APIs), batch processing, graphical user interface (GUI) for file upload, software development kits, and more
- Libraries for natural language processing, machine learning, neural networks, deep learning, and more
- Top programing languages for learning and model generation
- Developer portals and sandbox capabilities to verify and validate AI model efficacy
- Management information reporting (MI) platforms for regulatory reporting and analytical solutions
The next wave of intelligent banking
The open banking regulation has already blurred the lines between industries through its next avatar – open finance.
Banks have started gearing themselves for a cooperative competition; data or information is becoming their bridge. Open intelligence helps banks create a multitude of offerings for consumers, regulators, ecosystem players, and business operations. Banks can achieve a return on investment through business models like as-a-service, as-a platform, and more, along with contextually differentiated digital propositions, where human intelligence made a difference in the past. When employed in carefully controlled processes, open intelligence will help organizations realize exponential value from their businesses.
Contact
Contact
