The financial services industry, a leader in technological innovation, has seen artificial intelligence and generative AI hailed as transformative. However, adoption rates are often overhyped, with implementation stalling beyond a few pioneers. Despite a surge in proof-of-concept projects, evidence suggests a slowdown in full-scale AI deployment. Is there a reason?  And how can we solve the dilemma?

 

The reality behind the hype

Everest Group estimates most proof of concept (POC) pilots – approximately 90% – will not move into production in the near future, and some may never move into production. This limited completion raises questions about the practicality and immediate benefits of AI in real-world business settings. The problem is that the reality behind the hype is more complex and challenging than anticipated.

 

Barriers to AI integration and scalability

Businesses may face many challenges that can impede AI adoption in financial services:

  1. Regulatory challenges: Stringent regulatory environments complicate AI deployment, with compliance concerns related to data usage, consumer privacy and responsible AI. This hinders organizations from fully leveraging the power of AI technologies.
  2. Technical limitations: Despite advancements, AI technologies — particularly GenAI implementations — often need substantial customization for specific financial contexts such as insurance underwriting and risk management, and credit management in banking. The lack of standardization and the high cost of bespoke solutions deter widespread adoption.
  3. Cultural resistance: Organizational resistance to AI-driven changes is notable, especially in traditional banking institutions where skepticism toward replacing established human-driven processes with automated systems prevails.
  4. Responsible AI: Developing and implementing AI in an ethical, transparent and accountable manner is imperative. Ensuring AI systems do not perpetuate biases or make unjust decisions is critical, particularly in financial decisions affecting consumer finances and risk evaluation.
  5. Data and digital maturity: The level of data and digital maturity within an organization significantly impacts its ability to adopt and scale AI technologies. Many financial institutions struggle with legacy systems that are incompatible with modern AI applications, necessitating further modernization. Effective AI integration requires a robust digital infrastructure that is capable of handling large volumes of data efficiently and securely.

 

The surge in proofs of concept

Despite challenges, the financial services industry is actively piloting AI projects in various domains:

  • Back-office operations: Leading institutions are using GenAI to enhance efficiency in policy and claims management with document search, customer support, records management and business process automation. However, cultural resistance and reluctance to change long-established practices often prevent widespread scaling.
  • Chatbots and customer interaction: Many banks have introduced AI chatbots to improve customer service, yet these tools often function in limited capacities without replacing more complex human interactions.
  • Fraud detection systems: AI systems have been piloted to detect fraud patterns, but a high rate of false positives has hindered their operational deployment.
  • Legacy system modernization: GenAI can analyze and rewrite millions of lines of code to improve performance and reliability. However, the high-risk nature of these initiatives necessitates a cautious approach.

These complications highlight the gap between promising pilots and full-scale AI implementation.

Everest Group estimates most proof of concept (POC) pilots – approximately 90% – will not move into production in the near future, and some may never move into production.

Five ways to solve the dilemma

Based on our experience helping multiple financial institutions leverage AI and GenAI, we have identified five best practices essential for overcoming roadblocks and achieving effective business outcomes with rapid and long-lasting value:

  1. Start with easy-to-implement use cases.
    Select use cases that are easy to implement and provide quick value. This approach helps overcome initial cultural resistance and builds internal expertise. To facilitate rapid assessment of potential applications in banking and insurance, we have identified solutions to provide:
    • Frictionless experience
    • Next-gen FS innovation
    • Smart operations
    • Security, risk and compliance
    • Platform modernization
  2. Innovate specific processes and products.
    Evaluate how GenAI can innovate your specific processes and products. While many organizations begin with basic GenAI solutions, true competitive advantage comes from moving beyond the obvious. This requires advanced full-stack mastery and development tools to overcome any technical incompatibilities in integration limitations with existing systems. Our experience shows that these integration limitations are often overestimated and can be resolved with the right technology and domain expert teams.
  3. Plan for scaling from the start.
    Consider how you will scale your GenAI use cases into robust enterprise-class solutions. This is often the most underestimated part of projects, leading many applications to remain in the proof-of-concept phase. Success relies on proper data management, ML/Ops, model tuning and sometimes infrastructure accelerators. Scaling GenAI across multiple use cases requires making data AI-ready with modern architectures and easing data access through data fabric and data mesh strategies. Proven methodologies and tools support this journey, unlocking the future of banking and insurance.
  4. Prioritize compliance and security.
    Tackle compliance, business sovereignty and security risks right from the start. Due to strict regulations, AI security and responsibility cannot be afterthoughts. Fortunately, excellent frameworks are emerging. We apply these frameworks in our responsible AI consulting and cybersecurity offerings. For example, we recently launched AIsaac Cyber Mesh, a GenAI-powered managed detection and response management solution, which won the 2023 Infosec Security Product/Service of the Year award.
  5. Embrace an iterative process.
    GenAI technology evolves rapidly. Leaders differentiate by moving quickly and being agile. Enterprises often start with use cases for modernization and operations, followed by customer experience, trust and compliance; and finally, disruptive innovation. This progression moves from simple to more disruptive applications step-by-step.

 

Embracing AI and GenAI in the banking and insurance ecosystem

The financial services sector is at a crossroads with AI and GenAI. Despite their potential and optimistic analyst projections, businesses are slow to adopt these game-changers due to technical, regulatory, cultural, ethical and infrastructural challenges. Implementing best practices can overcome these hurdles. Our GenAI acceleration program successfully addresses these issues for large financial institutions.

In fact, Eviden’s Generative AI Acceleration program has helped businesses use and scale generative AI. Offering end-to-end GenAI consulting, fast-to-value solutions and modular accelerators, we are enabling tangible business value, competitive advantage and innovation. This is further supported by partnerships with hyperscalers, AI ISVs, and high-performance processing providers.

GenAI is transforming the world as we know it. The future of banking and insurance is here, just not widely scaled yet.

 

Related Articles

View all blogs