雅虎香港財經 A.I. has already helped 36% of financial services execs reduce costs by 10% or more, says an expert at Nvidia
Good morning.
An expert at Nvidia, a chipmaker that recently reached a $1 trillion market cap, says that banks are all for artificial intelligence.
Many financial services professionals have reported seeing an upside to A.I. when it comes to customer experience, according to Nvidia’s 2023 State of A.I. in Financial Services report. They’re betting on the technology to more accurately assess risk and create operational efficiencies, in addition to reducing cost.
A survey of 500 global financial services professionals found that 36% decreased annual costs by more than 10% by using A.I. applications. And almost half (46%) said it has improved customer experience. Many of Nvidia’s financial services clients are operationalizing hundreds of A.I. projects, according to the report.
When it comes to implementing advanced artificial intelligence, “banks are already using generative A.I. for document extraction within insurance and mortgage documents, and search and retrieval from internal knowledge bases,” Kevin Levitt, Nvidia’s global industry business development director for financial services, tells me.
Training A.I. models
Nvidia’s research found that one of the obstacles for financial services firms to achieving A.I. goals is insufficient data sizes for model training. The recommendation is the use of generative A.I. to produce accurate synthetic financial data used to train A.I. models.
An example? “Generative A.I. can unlock new opportunities across the financial services industry,” Levitt explains. “It could be used in ‘Know Your Customer' and anti-money laundering operations, as well as transaction fraud detection. Thieves constantly develop new techniques to steal identities and/or execute fraudulent transactions. But sample sizes for these new methods are small.”
He continues, “synthetic data from generative A.I. allows banks to create large sample sizes of these new threat patterns and can train models to detect them faster and more accurately. Ultimately, banks or credit card companies can create troves of accurate synthetic data to train A.I. models to be able to identify fraud in milliseconds. Because the data is synthetic, it also addresses concerns about data privacy, as well as regulatory and compliance guidance that prohibits certain data from being transferred outside of certain geographies or internally within companies.”
But it’s important to note that “financial firms will not rely on generative A.I. models trained on general internet data like most of today’s models,” he says. The firms will train their own foundation models using their own company data, allowing banks to build models that perform more accurately and deliver a more personalized experience to customers, Levitt says.
Most banks will want to have guardrails that prevent generative A.I. from engaging with inappropriate topics, and they’ll want to know if the data used to train the models contain any bias, he says. BloombergGPT, which was recently released and developed in collaboration with Nvidia, is an example of how financial firms will release company-specific generative A.I. platforms, Levitt says.
Does he think generative A.I. will eventually impact all bank functions? “Yes, generative A.I. does have the potential to impact virtually every function from underwriting, to risk assessment to customer service,” Levitt says. "A.I. models will be able to analyze thousands of data streams in real-time to glean market intelligence to create summarized research reports and deliver improved investment returns for investors and portfolio managers," he says.
The future of tech in financial services sounds bright.
Sheryl Estrada
sheryl.estrada@fortune.com
This story was originally featured on Fortune.com
More from Fortune:
5 side hustles where you may earn over $20,000 per year—all while working from home
Looking to make extra cash? This CD has a 5.15% APY right now
Buying a house? Here's how much to save
This is how much money you need to earn annually to comfortably buy a $600,000 home
