Morgan Stanley has unveiled its new internal AI model for research tasks
The AI model is built on ChatGPT software, with OpenAI releasing an enterprise tier in August
Morgan Stanley’s share price rose 0.4% on Monday
Are banking juggernauts tech companies now? That’s the latest question we’re pondering after Morgan Stanley confirmed it’s launched an internal AI assistant based on OpenAI tech. It’s the first out of the gates to launch a custom AI model, though the likes of JPMorgan, Citigroup and Goldman Sachs are hot on their heels.
The move is likely the first of many we’ll see from the big banks as the opportunity for generative AI grows, with some seriously large numbers being touted by research firms on the potential value added to the banking industry.
Let’s get into the details of what Morgan Stanley has created and the market reaction to the news.
What’s Morgan Stanley’s AI play?
Banking giant Morgan Stanley is officially getting on the generative AI hype with a new artificial intelligence-powered assistant. With the catchy name ‘AI @ Morgan Stanley Assistant’, the new tool is designed for its financial advisors and support staff to access over 100,000 research reports and documents.
The AI program aims to save staff time on administrative and research tasks concerning questions about markets, internal processes and recommendations so that the advisors can focus on their client base more.
The new tech has been built on OpenAI’s GPT-4 software, having first announced in March that it was working on an AI assistant with the buzzy new tech. JPMorgan and Goldman Sachs also have similar projects, but Morgan Stanley is the first out the door with an internal customized AI program.
In the memo to staff, first reported by CNBC, Morgan Stanley’s co-president Andy Saperstein said the new tool would “revolutionize client interactions, bring new efficiencies to advisor practices, and ultimately help free up time to do what you do best: serve your clients.”
Morgan Stanley also has more AI tools planned, including running a pilot on an AI program called Debrief that automatically summarizes client meetings and generates follow-up emails.
Has OpenAI’s enterprise tier been successful?
Morgan Stanley is just one of the many companies that have been an early tester of OpenAI’s bespoke AI programs, with the AI start-up introducing a separate enterprise tier for businesses in August.
The new tier is designed for businesses of varying sizes and industries and includes access to GPT-4 with no usage caps, faster performance and API credits. The pricing tier varies according to the enterprise client’s size and needs. OpenAI confirmed several companies were part of the beta process, including Block, Canva and Duolingo.
The most critical differentiation with the enterprise tier is that OpenAI’s model isn’t trained on the data the company submits. In a blog post, OpenAI confirmed, “We do not train on your business data or conversations, and our models don’t learn from your usage”. Instead, ChatGPT can be used by the client to train its own custom model for customer service, research and administrative tasks as some example use cases.
However, there are still risks over the new enterprise model. It’s not clear what training dataset is used to train ChatGPT-4 and whether it might involve copyrighted material in its usage. Hallucinations, where the AI model gets confused and hallucinates facts, are also a concern.
Are other banks looking at generative AI?
It’s fair to say the opportunity for banking and generative AI to partner together is massive. JPMorgan confirmed back in May that it was looking to develop a ChatGPT-like AI model to handpick investments for customers. The banking titan is also investing $1 billion in AI and data analytics in 2023, with a view to investing either the same or more every year.
That’s nothing compared to the potential return on investment. JPMorgan predicts it will see $1.5 billion in realized value for 2023 alone, while McKinsey estimates AI could create up to $1 trillion of additional value every year for the global banking sector.
Citigroup also laid out its AI plans recently, with CEO Jane Fraser confirming the bank had been working on generative AI models for the last three years. Fraser commented in her LinkedIn post that “the risks of not embracing generative AI far outweigh the risks of engaging with it.” Citi, along with JPMorgan and Goldman Sachs, had banned ChatGPT on their trading floors until a full risk assessment was completed.
The Evident AI Index ranks the top 23 banks in North America and Europe on how well-prepared they are for the coming AI revolution, with JPMorgan taking the top spot by some distance. Ranked on talent, innovation, leadership and transparency, U.S. banks are further ahead of their European counterparts and took up seven of the top 10 spots. Morgan Stanley held the tenth spot in the index.
What else is happening with Morgan Stanley?
The AI news coincided with the announcement on Monday that Morgan Stanley was being sued for at least $750 million by private equity firms that claim they were defrauded in a bad high-speed rail company investment.
Subsidiaries of Certares Management and Knighthead Capital Management have charged Morgan Stanley with contract violation and fraud, alleging that the firm improperly restructured a deal involving their investment in a loan to Brightline Holdings.
Brightline Holdings, also a defendant in the case and backed by private equity firm Fortress, operates a Florida rail system and plans to develop more railway lines between LA and Las Vegas.
The plaintiffs said Morgan Stanley convinced them to invest $281 million but concealed information about a Brightline preferred share deal that should have needed the loan to be paid at a “make whole” amount of roughly $750 million. The bank is also accused of forging documents.
Morgan Stanley said in a statement, “The firm does not believe the claims have merit and will defend itself vigorously.”
What was the market reaction?
Morgan Stanley’s share price closed 0.4% higher on Monday, possibly driven by the AI news. In some good news after the litigation bombshell, the stock has also risen 0.12% in premarket trading on Tuesday.
The bank’s share price is up 3.49% since the start of the year after a tumultuous March banking crisis left the banking sector‘s stocks struggling to recover. In comparison, rival JPMorgan has seen a 10.36% increase in its stock, but Goldman Sachs has declined 0.8% and Citigroup has lost 6.8% in value. SPDR S&P Bank ETF has seen just over a 16% decrease in its value in 2023.
The bottom line
Morgan Stanley has been the first to lay down the gauntlet and present a fully operational, internal AI model to make life easier for financial advisors and staff. We can imagine the other top banks will want to mimic the success soon with their own AI models.
As for Morgan Stanley, the bank saw a slight gain in its share price on Monday as a result. What’s interesting about AI in banking is that the potential value added pointed out by McKinsey is enormous – which is why so many banks are looking to hop on the generative AI bandwagon. Expect a flurry of banking-related AI announcements in its wake.