As the bank looks to outpace competitors in implementing artificial intelligence in the highly regulated sector, JPMorgan Chase & Co. is collaborating with US regulators and guiding them through its first batch of generative AI pilot projects to ensure all controls are in place.
Lori Beer, global chief information officer at JPMorgan, explained in an interview that “the goal is to assist regulators in comprehending how we develop generative AI models, how we manage them, and what are the new risk vectors”. She clarified that “it’s important to involve them early on and that it’s not just something we should think about, but also something they should think about. We share what we learn as we go along.”
Since ChatGPT’s launch a year ago, banks have been recruiting for AI-related roles and experimenting with generative AI, which can create emails, summarize documents, and generate intelligent responses for users. JPMorgan is recruiting and expanding faster than most. AI could be incorporated into “every single process” of the company’s operations, according to Chief Executive Officer Jamie Dimon, who referred to the technology as “extraordinary and groundbreaking” in his annual shareholder letter.
As per Beer, JPMorgan is presently testing artificial intelligence (AI) applications that have the capability to produce earnings summaries for each company the bank monitors. Additionally, the helpdesk service offers precise steps to solve problems rather than just sending customers to related articles to address an issue.
In an interview from the Innovation Lab conference room in a vast office park situated on the Outer Ring Road stretch in the suburbs of Bangalore, India’s technology hub, Beer stated, “We’re piloting, we’re learning, we’re figuring out.” We won’t be able to declare anything is in production until at least the first half of next year, based on what we’ve learned.”
JPMorgan has filed an application to trademark IndexGPT, which is listed as a product that gives customers investment advice. However, Beer clarified that this is just an application and not a product that is currently being developed. She stated that they want to protect their intellectual property and that it’s not something that will happen tomorrow. JPMorgan also developed a tool that analyses Federal Reserve speakers’ speeches to identify changes in policy and extract trading signals.
With a $15 billion yearly budget, Beer, who is headquartered in New York, is in charge of all the bank’s technological operations and 57,000 tech employees. India is home to one-third of the bank’s technologists, and she was there for a week. She used terms like “accelerate,” “transformative,” and “new paradigm” throughout the conversation, demonstrating her enthusiasm for artificial intelligence.
Banks have been investing billions in artificial intelligence (AI) systems to automate tasks like trading, risk management, fraud detection, and investment research for years. However banks are now pushed to create novel offerings due to the popularity of generative AI tools. McKinsey & Co Inc. projects that generative AI will bring up to $340 billion in value to the banking sector annually, including increased productivity.
Financial advisors at Morgan Stanley can now access the company’s database of approximately 100,000 reports during client conversations due to the implementation of an artificial intelligence tool. Software code is being written by developers at Goldman Sachs Group Inc. with the help of generative AI. And in order to comb through 1,089 pages of new capital regulations pertaining to the US banking industry, Citigroup Inc. turned to AI.
Beer stated that although generative AI makes it simpler to “build things quickly,” “the harder part is the validation and the controls.” As one of the first businesses to limit employee access to ChatGPT, JPMorgan is currently making efforts to guarantee that the sensitive data it handles in its pilot projects is well-protected. According to Beer, the industry is also devoting “a lot of R&D” to solving the issue of generative AI systems fabricating or hallucinating details. Additionally, it collaborates with regulators from the outset.
To get the right results on this one, Beer said, you have to see regulator guardrails. When it comes to controls, a globally systemically important bank will have different requirements than a startup.
