OpenAI logo on screen with the ChatGPT website on mobile, seen in this illustration in Brussels, Belgium, on December 12, 2022.
Jonathan Raa | Nurphoto | Getty Images
Morgan Stanley is rolling out an advanced chatbot powered by the latest technology from OpenAI to help the bank’s army of financial advisors, CNBC has learned.
The bank has tested the artificial intelligence tool with 300 advisors and plans to roll it out widely in the coming months, the bank said. Jeff MacMillanhead of analytics, data and innovation in the firm’s asset management division.
Morgan Stanley’s move is one of the first announcements from an established financial firm following the success of OpenAI’s ChatGPT, which went viral late last year by generating human-sounding responses to questions. The bank is a wealth management juggernaut with more than $4.2 trillion in client assets. The promise and dangers of artificial intelligence have been written about for years, but it wasn’t until ChatGPT that mainstream users understood the ramifications of the technology.
The idea behind the tool, which has been in development over the past year, is to help the bank’s approximately 16,000 advisers access the bank’s vast repository of research and data, McMillan said.
“People want to be as knowledgeable as the smartest person” in our company, McMillan said. “This is like having our chief strategy officer sitting next to you when you’re on the phone with a client.”
While generative AI has stunned users and sparked a race between technology giants to develop products, it has also brought down some users strange paths. Last month, Morgan Stanley analysts wrote that ChatGPT occasionally “hallucinates and can generate answers that are apparently convincing, but are actually wrong.”
User guardrails
Similar to ChatGPT, the tool directly answers questions for advisors. But it is based on GPT 4, which is a more advanced form of the technology that underpins ChatGPT.
And rather than the full content of the web, this tool only generates answers to the 100,000 or so studies Morgan Stanley has vetted for this use, which should reduce the number of errors. To further reduce accidents, the bank is having people check the accuracy of the answers, he said.
“We’re actually trying to break the platform” through human testing, he said. “With high-quality information, the better models and an ongoing monitoring process,” the bank is confident in its new tool, he said.
The Morgan Stanley logo can be seen in New York
Shannon Stapleton | Reuters
The move continues previous attempts by McMillan, including the 2018 introduction of machine learning algorithms that prompt advisors to contact clients or take other steps. With every new development, concern grows among knowledge workers that technology will one day be able to completely exclude humans.
“I think every industry is going to be disrupted in some way for what I’ll describe as routine, basic tasks,” McMillan said.
But machines can’t replace humans when it comes to catering to sophisticated customers, he said.
“These things don’t have any empathy; it’s just very clever math that’s capable of regurgitating knowledge,” he said.
