One of the biggest names in financial risk assessment has apparently deemed the tech behind ChatGPT safe enough for its own tools.
Moody’s Analytics is rolling out an AI-based chatbot that will help banks, insurers, and other clients more readily size up financial risks around business decisions. The agency built its new research assistant around Microsoft’s Azure OpenAI service, which lets businesses build their own language models for specialized purposes.
Tapping a tech prone to so-called “hallucinations” can be a tricky business in an industry that relies on timely and precise information. But there’s also an undeniable appeal in a system that can quickly parse reams of data, according to Cristina Pieretti, general manager of digital insights for Moody’s Analytics.
“We think the [commercial] potential is very big based on the reactions we’re seeing from customers,” Pieretti said.
Moody’s safeguards against inaccuracies by instructing the research assistant to cite sources for all the information it offers, and be up-front when it doesn’t have an answer. The tool is also designed to suggest related questions to help customers better home in on the answer they want.
“As an example, we don’t have stock prices in the rating agency…If you ask for the stock price of something, [the bot] is going to say, ‘I don’t know. But I know about the rating of this company,’” Pieretti said. “That creates a lot of confidence in the answers, not only because it’s limited to look at the Moody’s corpus of data, but also because it’s going to provide citations.”
Moody’s has also rolled out another tool called CoPilot to its more than 14,000 employees for internal use. In a pilot test group of clients using the research assistant, Moody’s claims the tool was able to reduce the time users needed for data collection by up to 80%.
Keep up with the innovative tech transforming business
Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.
“As opposed to you having to retrieve different parts of the puzzle, the research assistant is giving you the answer and telling you, ‘Here are the different parts of the puzzle that I used to give you that answer in case you want to dig deeper,’” Pieretti said.
The company eventually wants to expand the tool to cover new areas like climate risk, cybersecurity, and compliance, plus add new functions like deeper analysis and chart creation, according to Pieretti.
Moody’s is far from the only company that sees an opportunity in applying large language models to financial information. Bloomberg also trained a version of OpenAI’s model for a system called BloombergGPT. And banks are using the tech for tasks like generating financial summaries.
A McKinsey report last week found that “risk and legal” functions represented the biggest area of opportunity among all forms of AI in the banking world, with the potential to create $385 billion in value.
Moody’s sees the financial upside, too.
“When you tie the potential revenue effect with the efficiencies that you can see, I think that can give you a lot of insight into the size of the commercial opportunity,” Pieretti said.
Correction 12/13/23: This piece was updated to reflect that the research assistant was not for internal use at Moody's. The agency has a separate tool for employees called CoPilot.