JPMorgan Chase embraces AI first step: providing AI assistants for employees
JPMorgan Chase's AI plan consists of three stages: the first stage is to provide models to employees; the second stage is to enhance employee productivity by adding JPMorgan Chase's proprietary data; the third stage is to develop generative AI that can act as autonomous agents to perform complex tasks, making ordinary employees more like managers directing AI assistants
JPMorgan Chase has launched a generative AI assistant for its tens of thousands of employees.
According to media reports, the plan called LLM Suite is the initial stage of the bank's comprehensive promotion of AI technology, already serving over 60,000 employees to help them with tasks such as writing emails and reports. Insiders say that the software is expected to become as popular throughout the bank as the video conferencing program Zoom.
JPMorgan Chase did not develop the AI models itself, but designed a portal that allows users to access external large language models, initially using the LLM developed by OpenAI. Chief Data and Analytics Officer Teresa Heitsenrether expressed the hope to flexibly switch models based on different use cases, rather than relying on a specific vendor.
Since the launch of ChatGPT in late 2022, generative AI has rapidly penetrated major U.S. enterprises. Morgan Stanley has introduced two tools powered by OpenAI for its financial advisors. Apple has also integrated OpenAI models into its operating system, greatly expanding its application scope.
JPMorgan Chase CEO Jamie Dimon stated that this technology, known as the "cognitive revolution," may be as important as electricity, printing, and the internet, and is expected to improve the efficiency of almost all positions in the bank. As of June, JPMorgan Chase has around 313,000 employees.
Previously, JPMorgan Chase had restricted employees from using ChatGPT due to data security concerns, but has now found a way to both leverage the model and protect data.
LLM Suite has been widely used in JPMorgan Chase's consumer division, investment banking, and asset management business, helping employees write, summarize long documents, use Excel to solve problems, and generate ideas. JPMorgan Chase engineers can also directly integrate the functionality of external AI models into their programs using LLM Suite.
With over a decade of experience in traditional AI and machine learning, JPMorgan Chase's shift to generative AI was prompted by the emergence of ChatGPT. The application scope of generative AI far exceeds previous technologies, and JPMorgan Chase is testing various AI application scenarios and has already put some into production.
JPMorgan Chase is using generative AI to create marketing content for social media, plan itineraries for clients, and summarize meeting content. AI also helps consumer banks determine the locations of new branches and ATMs, and quickly find answers in call centers. In addition, AI prevents hundreds of millions of dollars in fraud in global payment operations every day.
Heitsenrether was promoted to head of the project in 2023, and she stated that the evolution of generative AI is divided into three stages:
The first stage is providing models to employees;
The second stage is adding JPMorgan Chase's proprietary data to enhance employee productivity;
The third stage is when generative AI can act as autonomous agents to perform complex tasks, making ordinary employees more like managers commanding AI assistants. This will change the industry composition, enhance the capabilities of certain employees, and may replace others.
Consulting firm Accenture believes that banking jobs are the most easily replaceable in all industries, with AI potentially adding $170 billion in profits to the industry within four years