Major financial institutions, including big brand banks, were among the many organizations revealing or launching generative AI projects in the months after ChatGPT’s launch. Of these projects, chatbots and internal assistants are a prevailing feature.
Morgan Stanley and its internal virtual assistant “made waves” in March 2023 when it was announced. The bot went live in September providing Morgan Stanley financial advisors with concise answers that can be used to quickly serve customers, from a database of 100,000 research reports and documents.
Chatbots in Banking – Across-the-Board Adoption and the Pivot to Generative AI
A US Consumer Financial Protection Bureau (CFPB) study published in mid-2023 found all the top ten U.S. commercial banks had deployed chatbots, and by 2022, 37% of banking customers had interacted with a bot. These, assumably rule-based, bots were limited technically, and customer complaints included wasted time and receiving inaccurate information. Still, the CFPB estimates that using chatbots instead of human agents saves $8 billion per annum or approximately $0.70 per customer interaction.
“Working with customers to resolve a problem or answer a question is an essential function for financial institutions – and the basis of relationship banking.”
The study details Capital One’s launch of its SMS chatbot Eno in 2017, which could check balances, review transactions, and help with payments. Furthermore, Bank of America’s chatbot Erica was used by 32 million customers in over 1 billion interactions between its launch in 2018 and October 2022.
The CFPB notes that the banking industry has begun to adopt generative AI chatbots or large language models (LLMs), having progressed over the years from local branches to contact centers, then mobile applications and on to live chat and chatbots.
A McKinsey customer care report for 2024 says 80% of respondents in its recent study across multiple industries are already investing in generative AI or plan to do so in the coming months. It adds that a European subsidiary of a global bank replaced its rule-based customer chatbot with a generative AI version and found it was 20% more effective at answering customer queries. The bank has further identified improvements hoping to double its performance increase.
Internal Assistants & Tools for Finance Professionals
Goldman Sachs
The CIO of Goldman Sachs suggested in April 2023 that its engineering staff work on a model to help employees store knowledge and answer customer questions on demand.
(Incidentally, Goldman Sachs Research analysts predict that tools using natural language processing could drive a “7% (or almost $7 trillion) increase in global GDP and lift productivity growth by 1.5 percentage points over a 10-year period”).
By November 2023, Reuters revealed that Goldman Sachs was working on no less than twelve generative AI projects for its business practices. Its most mature projects include a tool to write code in English language commands and to generate documentation. However, none of the projects are client-facing. George Lee, co-head of Goldman Sach’s office of applied innovation, says this is because of the “regulated nature of financial services” and that they were “moving very deliberately, very carefully, very thoughtfully.” Lee also added that a “human in the loop” is required for AI, to manage and to intervene where necessary.
Discover What is Human-in-the-Loop (HITL), and Why Does it Matter?
Morgan Stanley
The “Morgan Stanley Assistant” went “fully live” for the company’s financial advisors and support staff in September 2023, providing instant access to the information contained in around 100,000 research reports and documents. The assistant is conversation-based. Morgan Stanley Advisors must pose questions in full sentences as if speaking to a human, per CNBC, instead of using keywords.
Morgan Stanley co-President Andy Saperstein, referring to the assistant’s use by advisors, said at the time:
“Generative AI will revolutionize client interactions, bring new efficiencies to advisor practices, and ultimately help free up time to do what you do best: serve your clients.”
The bank’s head of analytics, data, and innovation, Jeff McMillan, said Morgan Stanley was the first major Wall Street firm to provide a bespoke solution based on ChatGPT-4 to employees. The assistant gives fast access to “intellectual capital,” saving advisors and customer service staff time when answering customer questions about markets recommendations and internal processes.
The assistant won’t be Morgan Stanley’s last AI tool. It’s also piloting a tool called “Debrief” that summarizes client meetings and creates follow-up emails.
Other banks are exploring AI…
JPMorgan Chase is working on software that uses AI to select investments. It was also revealed in November 2023 that the company was testing AI applications able to generate earnings summaries as well as a helpdesk service that provides “exact problem-solving steps,” per Bloomberg.
The bank is reportedly “walking” regulators through its AI pilots so they can see the controls in place. CIO Lori Beer told Bloomberg:
“It’s about helping regulators understand how we build the generative AI models, how we control them, what are the new vectors of risk.”
JPMorgan has applied to trademark its own technology, called “IndexGPT,” for selecting investments, but at the end of 2023, it said it wasn’t yet a product development. That was likely to progress in the first half of 2024.
RBC is also said to be working on a proof of concept that could be launched later in 2024 for an LLM that pulls data from policies and procedures for bankers when advising clients.
Banks and financial institutions can be hesitant to launch customer-facing generative AI products largely because of the associated risks of AI, and additional risks exacerbated by the sensitivity and impact of the sector. However, ING, in collaboration with McKinsey, built and launched a bespoke customer-facing chatbot that uses AI but with guardrails that, for example, ensure the bot avoids giving advice on mortgages and investment products. McKinsey notes:
“The speed with which the chatbot was built and deployed far outpaced the timeline that’s been required to develop previous industry-standard chatbots, which can take several years of programming and finetuning to get right.
What’s more, within the first seven weeks of use, the new chatbot was offering a better customer experience – helping 20 percent more customers avoid long wait times and offering instant gratification – compared to the previous solution.”
Discover further Generative AI Use Cases in Business – Lemonade Masters AI for Insurance
Build Your Own Internal Knowledge Repository in Minutes with CustomGPT.ai
Financial institutions appear keen to explore the significant benefits of internal use cases for AI, which allow a safer route than customer-facing AI applications while also enabling early adoption.
CustomGPT.ai, using advanced large language models (LLMs), allows any organization to build an internal employee-facing chatbot assistant simply in minutes with a zero-code experience. It’s easy to train a CustomGPT chatbot by uploading company documents such as HR guides or sales literature.
CustomGPT.ai accepts thousands of documents supporting over 1400 file formats. Internal chatbots can be complete or specific knowledge repositories, tools for HR, training, and employee onboarding, or assistants that sales teams can use to retrieve information quickly in real-time while interacting with customers.
Learn how CustomGPT.ai enhances customer support efficiency with instant, personalized responses and seamless communication powered by your business content.
