Banks spent around $3.3 billion in 2018 on a wide variety of AI applications, most of which centered on automated fraud detection and prevention, compliance analysis and investigation, and program advisors.
AI’s expanse is wide-ranging across the banking value spectrum — from enabling improved data-driven decisions using machine learning (ML) in the back office, to applying human perception in improving automation in the middle office, to directing human interactions in the front office to drive a better customer experience (see Figure 1).
Source: Cognizant Analysis
One of the primary focus areas for AI experimentation in the banking industry is to augment customer experience with applications such as personalized voice assistants or chat-bot-powered transaction banking and payments. With 92 million millennials coming of age in the U.S. alone, customer experience is becoming a major differentiator.
ML-enabled regtech and risk decision support is the other key area of AI investments at banks. These investments are aligned closely to Digital 1.0, which harnessed big data and enabled data-driven decision making through analytics. With Digital 2.0, ML-led AI solutions are possibly the only way to parse and interpret the vast amounts of data created, stored, and accessed by banks in the last decade.
Of course, the journey of value discovery for AI-led solutions and the marginal benefits of investment are still evolving. Nevertheless, as with digital adoption, laggards will pay a price in terms of both revenue opportunities and costs.
In Part 2 of this two-part series, we look at ways banks can successfully apply AI.
Leveraging AI in banking
As illustrated in Figure 1, opportunities abound for AI in the banking industry. However, in our experience, adoption is taking many forms and approaches. The common ones include proofs of concept and pocketed experiments with the help of tech vendors and AI service providers, as well as investments and joint ventures with fintechs and startups working on AI-enabled tech. A multi-industry survey by Cognizant found that financial services firms identified virtual agents (72%) and analysis of natural language (56%) as the most commonly used technologies.
In a 2018 survey by the Financial Times of leading banks, more than two-thirds of respondents had investments or joint ventures related to AI adoption, yet only a small minority had board-level representation with responsibility for this adoption. In order to unleash the potential of AI, banks need to take a holistic and strategic view. This is especially true in light of the unique challenges of managing and governing AI, with its broad applicability across the value chain and its ability to completely transform tomorrow’s workplace and workforce.
Banks can choose from a spectrum of models, ranging from highly centralized centers of excellence (CoEs) to the completely distributed business unit-driven AI pods approach. While we do not recommend a prescriptive AI operating model, it is imperative that banks make a strategic decision weighing the pros and cons of each approach, instead of letting organic processes make the decision for them.
Centralized approaches such as CoEs and consulting models provide for stronger AI governance with better alignment to business goals, thereby bolstering inter-firm sharing of knowledge through shared data models, algorithms and libraries. However, a more functional, business-driven AI structure could allow for more freedom in value discovery in a field that is still maturing and calls for deep operational knowledge and expertise to advance. Marrying the two models to maximize AI traction will require executive leadership, targeted vision, and understanding of the potential from both a bottom-up and top-down perspective.
AI adoption requires a multi-disciplinary approach that combines expertise from technology specialists across the AI sub-fields, banking domain and management so that banks can:
Develop an enterprise-wide AI vision and roadmap based on an understanding of both the current limitations and disruptive potential.
Advance an AI operating model and governance with a view toward aligning AI adoption with the bank’s stated values and policies, as well as regulatory requirements.
Provide design and solution guidance to AI implementation using best practices.
Build ashared library of AI services to speed up implementation and reduce rework.
Plot the AI pipeline that is in line with strategic goals, business unit needs and tech maturity.
Emphasize organization change management and a communication strategy for the success of the AI to get out ahead of a workforce change, cultural shifts and the need for deeper cross-enterprise collaboration.
The banking industry is already feeling AI’s effect throughout its value chain. What makes AI unique is that it introduces a new partner in the workforce that must be effectively managed — in ways similar to what humans require.
While the emergence of a general, pervasive AI is still over the horizon, we believe a combination of mature and narrow AI working alongside humans can bring major improvements in productivity that will unlock transformational business models and uncap novel revenue sources.
To leverage these benefits, banks must make strategic choices on how they embrace AI adoption, with full cognizance of their underlying needs. We believe success will come only with a clear leadership vision supported by a focus on operating model readiness that balances both innovation and governance.
Ardhendu Acharya, Senior Manager Consulting, Kumar Ray, Manager Consulting,Amit Anand, Assistant VP, and Madhu Ponnuveetil, Senior Director, at Cognizant’s Banking & Financial Services Consulting practice contributed to this article.