Alejandro Mijares
Founder and Chief Executive Officer, Mijares Consulting
EXECUTIVE SUMMARY
The adoption of Artificial Intelligence (AI) within community institutions is marked by significant challenges, as highlighted by the low response rate (28% of 61 responses reviewed) to a survey question on AI implementation challenges. These challenges span across various critical areas:
· Governance issues arise due to the lack of clear strategies and policies tailored for AI
· Resource limitations, particularly in financial and technical expertise, hinder the effective implementation of AI
· Operational Integration of AI into existing systems proves complex
· Regulatory challenges add layers of compliance difficulties
In the ongoing exploration of Artificial Intelligence’s (AI) multifaceted role within the financial sector, we are excited to introduce Part 3A of our comprehensive series. This segment delves into connecting the insights drawn from various agencies’ Requests for Information (RFI) with the detailed analyses conducted by our Firm. Previously, in Part 1, we examined the proactive measures undertaken by federal agencies to gather information. This entailed soliciting perspectives on AI applications in financial institutions and delineating the structure and purpose of the RFI. In Part 2, we pivoted our attention to an exhaustive evaluation of responses received from a diverse array of 61 organizations and individuals. These responses, addressing 17 pivotal questions from the RFI, offered a rich spectrum of viewpoints and insights, enriching our understanding of AI’s current and potential impact in the financial realm.
In part 3, AI Adoption in Community Institutions (Question 9), a mere 28% of organizations provided answers, marking it as the question with the lowest response rate, acknowledged facing challenges in implementing AI within community institutions. This statistic may suggest a perception that AI’s impact is less pronounced in smaller, community-focused financial entities, or it could indicate a lower level of engagement with AI technologies in these institutions. Community institutions encounter a distinct set of challenges in adopting AI compared to their larger counterparts. These challenges often include limited budgets, a scarcity of specialized expertise, and heightened concerns regarding data privacy and security. It is imperative for policymakers, regulators, and industry leaders to identify and promote practices that address these unique challenges. Doing so is essential to facilitate a more seamless integration of AI technologies in community institutions, ensuring their active participation in the ongoing technological revolution and preventing a digital divide in the financial sector. We noted the following:
AI Use by Community Institutions (Question 9 of RFI):
Question 9 (a): Do community institutions face particular challenges in developing, adopting, and using AI?
Question 9 (b): If so, please provide detail about such challenges. What practices are employed to address those impediments or challenges?
The utilization of Artificial Intelligence (AI) by community institutions presents a unique landscape, as evidenced by the fact that only 28% of organizations, the lowest response rate for the survey, acknowledged encountering challenges with AI implementation. This statistic might suggest that AI’s impact is perceived as less critical in smaller, community-focused institutions, or it could indicate a lower level of engagement with AI technologies in these entities. Unlike their larger counterparts, community institutions face distinct hurdles in adopting AI, which may include, but not limited to, resources limitations, a deficit in specialized expertise, or apprehensions regarding data privacy and security. It is essential for policymakers, regulators, and industry leaders to recognize and address these specific challenges. Developing and promoting practices that cater to the unique needs of community institutions is crucial. This approach will not only facilitate smoother AI adoption but also ensure that these smaller entities are not left behind in the ongoing technological revolution.
Our analysis of the responses reveals that community institutions encounter a spectrum of challenges in developing, adopting, and using AI. These challenges span across various critical areas: Governance issues arise due to the lack of clear strategies and policies tailored for AI; Resource limitations, particularly in financial and technical expertise, hinder the effective implementation of AI; Operational Integration of AI into existing systems proves complex; Regulatory challenges add layers of compliance difficulties; Third-party oversight is necessary yet challenging due to dependency on external vendors; Data Management and Data Security concerns are paramount, given the sensitivity and volume of data AI systems require; and Testing & Validation of AI models is a significant hurdle due to the lack of resources and expertise. Each of these areas presents unique obstacles that community institutions must navigate to successfully leverage AI technologies, underscoring the need for tailored solutions and support to address these multifaceted challenges. In this article we will cover challenges in Governance, Resource Limitations, Operational Integrity, and Regulatory Challenges. In the next article we will cover Third-Party Oversight, Data Management, Data Security, and Testing & Validation. The following is the detail analysis performed by the Firm:
Governance
Community financial institutions encounter specific challenges in the governance of developing, adopting, and using Artificial Intelligence (AI), which can significantly impact their ability to leverage these technologies effectively. These challenges often stem from the following:
· Lack of Policies and Frameworks
· Inadequate Business and IT Strategy Alignment
· Deficiency in Specialized Committees
· Insufficient Risk Management Practices
· Limited Expertise in AI Governance
· Challenges in Stakeholder Engagement
· Lack of clear direction and resources
The most fundamental issue is not knowing where to start, especially for smaller institutions with limited funding and resources compared to larger financial entities. These institutions recognize the importance of AI/ML but often lack in-house development capabilities and may rely on third-party providers, developers, and consultants. As such, community banks are almost incapable of evaluating AI/ML technology without an independent framework for assessing models’ soundness (whether the model is consistent, reliable, and unbiased).
Leadership plays a crucial role in driving a top-down AI business strategy, which is often lacking in these institutions. Additionally, the management of adoption and engagement presents a challenge. Institutions might acquire AI/ML tools to fulfill a perceived need but lack a comprehensive strategy for their effective use, leading to underutilization.
Moreover, community institutions struggle with accessing AI technology benefits due to uncertainties about their technical expertise, regulatory compliance, and resource limitations. They often express concerns about meeting technical requirements for model documentation and rely heavily on third-party expertise for oversight and monitoring. Smaller institutions, in particular, face challenges due to limited in-house machine learning knowledge, essential for implementing effective model governance. This situation underscores the need for clear strategies, adequate resources, and knowledgeable leadership to harness the potential of AI and ML in community financial institutions.
Resources Limitations
Community financial institutions often find it challenging to fully leverage the advantages of AI technology, primarily due to uncertainties surrounding their technical capabilities, regulatory compliance responsibilities, high fixed costs associated with each adopted model’s due diligence (absolute efficiencies benefit for a community institution is usually smaller due to lower business volume which makes it harder to justify the large fixed overhead needed) rapidly evolving nature of AI, and perceived shortcomings in expertise, staffing, and financial resources. This lack of institutional knowledge means that they also struggle to evaluate AI/ML models and tools for both efficacy and soundness and to understand which AI/ML solutions are suitable for their operations, how these technologies function, or the controls needed to mitigate risks. Consequently, these institutions often resort to relying on third-party providers, which introduces additional risk mitigation considerations.
These institutions frequently voice concerns regarding their capacity to fulfill the stringent technical requirements for model documentation, as stipulated by supervisory authorities and regulators. This apprehension is compounded by doubts about their ability to rely on third-party expertise for the necessary monitoring and oversight. Such uncertainties contribute significantly to the difficulties these institutions encounter in effectively adopting and utilizing AI technologies. This situation underscores the need for enhanced support systems, clearer regulatory guidelines, and more accessible resources to aid these institutions in navigating the complexities of AI implementation and management.
The successful development, implementation, and application of AI are contingent upon an organization’s capacity to secure top-tier AI talent, harness sophisticated analytical tools, and establish a robust data and analytics infrastructure. This technology and expertise gap expands beyond the realm of data scientists, encompassing roles such as business analysts, marketers, and loan officers. These professionals need to understand how to effectively utilize the results generated by AI systems. Community institutions can bridge this gap by leveraging the techniques of federated learning. Federated learning is a method that allows AI models to be trained on multiple devices without the need for data transfer. This ensures data privacy and security, while still contributing to the AI model’s training. In pursuit of this, these institutions could form a consortium with other like-minded organizations. Alternatively, they may opt to engage a third party to develop and manage their models. This would allow them to focus on operational aspects and monitoring processes, ensuring efficient use of resources and expertise.
Operational integration
A challenge that arises is achieving operational integration to drive genuine transformation in products and offerings. The accessibility of technology emerges as a less conspicuous hurdle. Smaller organizations, reliant on third parties for their technological infrastructure, may find themselves at a disadvantage. Given that the transformative power of AI is anticipated to be most influential in areas related to product development, these smaller entities might lack the capacity to conduct the necessary research and development required for AI implementation.
Additionally, the management of adoption and engagement presents a challenge. Institutions might acquire AI/ML tools to fulfill a perceived need but lack a comprehensive strategy for their effective use, leading to underutilization. To optimize organizational learning that capitalizes on the strengths of both machines and humans, processes need to be adapted accordingly. Smaller organizations might find themselves at an advantage in this regard. With typically less bureaucracy, they often face fewer obstacles in terms of organizational resistance and can more readily adjust and enhance their processes to seamlessly integrate AI into their daily operations.
Additionally, the economies of scale that come with operating these processes for a multitude of models present a significant challenge for smaller financial institutions to attain. This deficiency in institutional knowledge results in these institutions facing challenges in effectively assessing AI/ML models and tools, not only in terms of their efficacy but also their soundness and reliability. The absolute efficiency gains are typically smaller due to their lower business volume. This makes it more challenging to justify the significant fixed overhead required to conduct due diligence for each model adoption.
Regulatory challenges
Regulatory hurdles emerge from the nascent nature of AI, a field where laws and regulations have not yet fully evolved at the time this article was written. FIs desiring to implement AI need to navigate through a regulatory gray area, as there is no clear directive on permissible AI activities or how these activities will be evaluated by examiners.
Additionally, most examiners, while adept at traditional banking operations, possess only a beginning understanding of AI. In the absence of regulatory guidance for approving or measuring AI technologies, they often struggle to assess an FI’s use of AI, particularly at smaller institutions. This scenario puts FIs at a disadvantage when trying to justify the appropriateness of an AI model for their operations or in demonstrating its non-discriminatory nature.
The discomfort of examiners in approving a technology they are starting to understand could result in FIs retracting significant investments in AI technologies. Despite due diligence and transparency, this hurdle is insurmountable. Examiners, trained to review explicit formulaic methodologies, are cautious not to make mistakes, thereby curbing innovation. The deterministic yet unexplainable nature of AI poses a significant challenge to existing examination models.
Conclusion
Part 3A of our series on AI Adoption in Community Institutions sheds light on the nuanced and complex landscape of integrating Artificial Intelligence within the financial sector, specifically focusing on community institutions. Despite the potential benefits, these entities face a unique set of challenges that hinder their ability to fully embrace and utilize AI technologies. Limited response rates to our survey indicate a perceived lower impact or engagement with AI, underscoring the need for a nuanced understanding of the technology’s role in smaller, community-focused institutions. The challenges highlighted—ranging from governance issues, resource limitations, operational integration difficulties, to regulatory uncertainties—reflect the multifaceted hurdles that community institutions must navigate.
The analysis underscores the importance of tailored support and strategies to overcome these barriers, including the need for clear policies and frameworks, alignment of business and IT strategies, specialized committees, sufficient risk management practices, and expertise in AI governance. Furthermore, addressing resource limitations, enhancing operational integration, and navigating regulatory landscapes are critical for the successful adoption and utilization of AI.
The findings call for a collective effort from policymakers, regulators, and industry leaders to facilitate a more inclusive and supportive environment for AI adoption in community institutions. By acknowledging and addressing these specific challenges, there’s an opportunity to ensure that community institutions are not left behind in the technological revolution, thereby preventing a digital divide in the financial sector. Ultimately, fostering a supportive ecosystem for AI integration in community institutions will enable them to harness the potential of AI, contributing to innovation, efficiency, and enhanced service delivery in the financial industry.
Join the AI Revolution: Transforming Community Banking with AI – Engage with Us on LinkedIn
We invite you to read Part 3 B, which offers an in-depth look at the use of AI in finance, linking insights from the agencies’ RFI with our firm’s findings based on the analysis performed by reviewing 61 responses to the RFI.
As we conclude this insightful exploration into the challenges and opportunities of AI in the financial sector, we invite you to stay engaged with this crucial conversation. If you found value in this article and believe others in your network could benefit from these insights, please consider ‘Liking’ and ‘Sharing’ it. Your engagement helps broaden the discussion and brings diverse perspectives to the forefront.
Don’t forget to ‘Follow’ for more updates and in-depth analyses on AI and its evolving role in finance. Your involvement is not just appreciated – it is essential in shaping a well-informed community around these pivotal technological advancements. Let’s continue to learn, share, and grow together in this dynamic field.
Thank you for your support and engagement!
