FINDING A COMPETITIVE EDGE WITH AI
The use of artificial intelligence (AI) in banking is not new, but leveraging it fully still offers an important competitive edge for established firms. In a recent survey by digital consulting firm Genpact, 97% of banks report using AI today, but only 19% say that they are using it across the whole banking value stream . With industry-wide cost savings from AI estimated at $447B by 2023, the time is now for banks to invest in integrated AI strategies that extends across business lines and functions, and emphasizes partnerships and hiring top talent .
CREATING VALUE FOR CLIENTS, THE FIRM, AND THE INDUSTRY
JPMorgan Chase (JPMC) is leading the way in holistic AI adoption. The bank has distinguished itself by its level of investment, aggressive hiring, and comprehensive approach to implementing and managing AI across the firm. JPMC’s efforts in AI can be divided into applied initiatives and cutting-edge research . The firm’s research agenda has four foci that have and will continue to shape new AI products and services for JPMC and the banking industry as a whole :
- Data & Cryptology – New ways to clean, integrate and generate data needed to train machine learning (ML) algorithms
- Learning From Experience – Exploring boundaries of ML through deep learning & reinforcement learning techniques
- Explainability & Interpretability – Creating techniques and algorithms that are understandable to regulators and clients
- Ethics & Fairness – Developing unbiased and ethical models to build and maintain trust
JPMC is using its research findings in six applied AI initiatives, which aim to reduce costs and drive additional revenue for the firm. Below is an overview of each initiative bucketed by the primary way it creates value :
- Anomaly Detection – Incorporating machine learning into fraud detection processes helps JPMC minimize and mitigate risks for clients and the company as a whole 
- Intelligent Pricing – Augmenting traditional pricing models with predictive analytics enables more accurate risk assessment and reduces losses
- Smart Documents – Solutions such as Contract Intelligence (COiN) identify meaningful information from lengthy text sources to significantly reduce contract review time and cost 
- News Analytics – Predictive analytics and natural language processing used in tools such as the Emerging Opportunities Engine help aggregate information from multiple sources to help clients make rapid, smart investment decisions 
- Quantitative Client Intelligence – Analysis of client communications across multiple channels allows JPMC to provide better customer services and customized product recommendations
- Virtual Assistants – Virtual assistants and chat bots are used to decrease time-to-response for client and employee requests, thus improving client service and operational efficiency
FACTORS ENABLING SUCCESS
JPMC’s success in AI is made possible by its strong financial commitment and coordination across internal and external partners. With an $11.4B technology budget in 2019, the firm is able to support a staff of ~53,000 technologists led by AI superstars such as Manuela Veloso, the former head of machine learning at Carnegie Mellon, and Apoorv Saxena, who previously led product management for cloud-based AI services at Google . An AI Center for Excellence and an AI/ML Council are two organizational features that coordinate efforts across all these people and ensure that learnings from the technology side of the business permeate all other areas of the firm . JPMC has also invested in outside partners to accelerate the development of new products and services; the most notable include working on a rapid, online small business loan platform with OnDeck, and partnering with Roostify to launch a self-servicing mortgage platform [ .
CHALLENGES & OPPORTUNITIES
Despite making AI a strategic focus, JPMC continues to face deployment challenges related to legacy systems and processes, talent acquisition, and external resistance. As with other incumbents, existing systems likely take up a majority of JPMC’s technology budget and integration with or circumvention of these systems slows down AI implementation . Additionally, full AI integration requires recruiting hard-to-find technology workers and getting buy-in from existing employees whose jobs will be impacted by the technology. While the company has mechanisms in place to recruit technologists and to disseminate AI information, they will continue to have to fight for talent from Silicon Valley tech firms and work to assuage the fears of current employees around job displacement. The final challenge JPMC faces is from external pressures, namely gaining clients’ trust in the security and efficacy of AI, and explaining/justifying their use of AI to regulators .
Opportunities abound for JPMC to further improve its AI products and services and to address the challenges described above. The firm should learn from the successes of its competitors to better its AI portfolio (e.g. Bank of America’s lauded virtual assistant Erica, PNC’s data infrastructure overhaul to make information more useable for AI/ML, and Wells Fargo’s predictive banking features for retail customers) . By continuing to improve products and focusing research that makes AI simpler and more explainable, JPMC will find less resistance from clients and regulators going forward. Finally, JPMC should double-down on its investment to reskill – and UPskill – its existing employees so they are more capable, receptive, and empowered to move rather than leave the company as AI solutions become more prevalent .