BofA’s Tech Chief Says AI Delivers Returns for the Bank

Bank of America's AI Strategy: A Deep Dive

Key Concepts: Digital transformation, self-service, client satisfaction, AI adoption, generative AI, large language models (LLMs), productivity, coding agents, ROI, model agnostic approach, business process optimization.

Digital Transformation and Self-Service

Bank of America has undergone a significant digital transformation, resulting in:

• 99% of customer interactions being self-service digital.
• 65% of account openings being done digitally, compared to 0% ten years ago.
• Client satisfaction scores increasing from the high 70s to 89.1.

This shift towards self-service channels and automation has yielded positive results in client satisfaction.

Erika: The AI-Powered Virtual Assistant

• Erika is an AI-powered virtual assistant that allows customers to interact with the bank using natural language.
• It simplifies the interaction process by allowing customers to express their needs in their own words, rather than navigating complex systems.
• Erika has facilitated 3 billion transactions and continues to grow exponentially.
• Initially designed for customer interactions, Erika's capabilities have been extended to internal use for employees.

Example: A high net worth advisor can use Erika to summarize complex trust policies and prepare for client meetings. Junior bankers can use generative models to create initial drafts of client dossiers, saving hours of preparation time.

AI Adoption and Employee Empowerment

• Bank of America has over 1400 patents and 250 AI models in production, including fraud detection models.
• The company is using generative AI to augment employee capabilities and improve productivity.
• 17,000 developers use coding agents to assist with their day-to-day activities.
• The use of coding agents allows developers to focus on more innovative work by automating mundane tasks.

Measurement: Productivity gains are measured using metrics such as cost per story point and defect density.

Multilevel Approach to Generative AI

Bank of America adopts a multilevel approach to generative AI:

1. Embrace Commodity: Utilize foundation models developed by others.
2. Model Agnostic: Avoid being tied to a single solution.
3. Build on Top: Develop complex workflows and business processes on top of foundation models.

The company is intentionally model agnostic and explores various options.

Rationale: Foundation models provide a base upon which to build specialized applications.

Business Process Optimization

• The key to unlocking productivity with generative AI is to apply it to business processes.
• This involves understanding current processes, identifying areas for simplification and elimination, and then using AI to automate and improve them.
• Bank of America measures the ROI of AI deployments by tracking metrics before and after implementation.
• The company emphasizes the importance of failing fast if a process does not lend itself to AI automation.

Quote: "The way you unleash productivity is applying these to your business process, understanding how your current process work. Don't just try to replicate it. Figure out where you can eliminate parts of it, simplify it, and then use it to go after it."

Impact on Employees

• The adoption of AI has not led to a reduction in the number of employees.
• Instead, it has allowed the company to address a backlog of work and modernize existing systems.
• Employees have generally reacted positively to being augmented by AI, as it frees them to focus on more innovative and strategic tasks.

Quote: "Our first reaction from our teammates has been this frees them to do more innovative work rather than the mundane work that was consuming their time and resources."

Conclusion

Bank of America's AI strategy is focused on leveraging both internal and external resources to drive digital transformation, improve client satisfaction, and empower employees. The company's multilevel approach to generative AI, combined with a focus on business process optimization and ROI measurement, positions it to effectively utilize AI to achieve its strategic goals. The emphasis on augmenting, rather than replacing, employees is a key factor in the successful adoption of AI within the organization.