Teslar Insights Report - Digital - Flipbook - Page 3
SPONSORED
Banks’ Generative AI Plans
for 2024
AI Use Cases Explored by Banks
Source: Cornerstone Advisors’ 2024
Technolog y Survey
Source: Bank Director’s 2024
automation within the institution, says
Chief Information Officer Brad Miller.
What’s Going on In Banking report
53%
of respondents had
discussed generative AI
at the board or executive
team level.
27%
said it’s not on their
radar.
14%
planned to invest and/or
implement in 2024.
6%
had already invested in
or deployed generative AI.
CONTENT
Generative AI assists with internal policy
Fraud detection or prevention
84%
generation and with light scripting and
coding within his team — areas that help
Customer service
his staff be more efficient without need-
67%
ing to hire more people.
The $1 billion bank also uses an AI-
Back-office efficiencies
67%
driven fraud tool to identify anomalous
transactions and risky customer behavior;
Targeted marketing efforts
it can even pause transactions for manual
61%
Credit underwriting
30%
staff review. The efficiency gains from
these tools could “easily pay for at least
a full-time employee a year — probably
more at this point,” he says. The bank
AI use cases had not yet been identified
5%
hasn’t deployed AI in its credit function.
Legacy National leverages vendors
for its AI tools, which is how most banks
will enable or add this technology. In
the credit function, banks could leverage existing loan origination software or
memorandums used in debt collections.
Ehrhardt says generative AI could be
if the review reveals borrower changes
credit monitoring tools that already have
outside of an acceptable risk range.
governance and due diligence around bank
great at processing reams of data and
customer information as part of their ven-
summarizing facts in a digestible narrative. The technology can combine
structured data sets — like financial state-
dor relationship. Evident AI recommends
Deployment Considerations
Bank regulators have flagged genera-
that banks make sure their vendors apply
emerging AI-specific standards and frame-
ments and cash flow — with unstructured
tive AI’s ability to create inaccurate or
works, and review their current software
external data. An AI tool could make sure
fabricated results.
evaluation, selection and procurement
the business is properly registered and
Before deploying a new generative AI
processes for AI-specific addendums.
legitimate. If the borrower is a restaurant,
tool, executives should consider potential
the tech could analyze customer reviews;
applications and parameters for the tech-
credit specifically, Ehrhardt adds that ex-
if it’s a commercial real estate property, it
nology, prospective vendors they would
ecutives will still need to set guidelines on
could analyze the location, rent rolls and
work with and the data the tool would
what external data AI should incorporate
vacancy rates.
analyze.
into its analysis. What directories should
“There are many areas in lending
Bankers and any AI users “have an ob-
For generative AI use cases within
it search to verify a business’ legitimacy?
where we create stories and memos to
ligation to make sure AI-generated data is
Should an AI tool scrape LinkedIn profiles?
describe situations, largely based off the
accurate, and it’s not made up,” Ehrhardt
Should it include pictures of the loan ap-
data and facts the bank has available,”
says. A generative AI tool used by credit
plicant on vacation, flagging concerns of
Ehrhardt says. “That’s a great spot for
analysts would aggregate, analyze and
embezzlement or inappropriate spending?
generative AI.”
summarize internal bank data about the
Ehrhardt believes generative AI ap-
A memo written with generative AI
borrower, along with external informa-
plications could become “commonplace”
could highlight the strengths and weak-
tion from sources the bank has deemed
in financial services firms in the next
nesses of the potential loan, assisting the
acceptable and relevant. “There’s so much
few years. That promises to reduce the
credit team while giving them the final
data available,” he says.
time and effort associated with creating
say in approving or denying a loan. A tool
Legacy National Bank, the bank unit
internal credit documents while main-
used for annual credit reviews could be
of Springdale, Arkansas-based Legacy
taining or increasing accuracy to speed up
programmed to alert credit employees
BancShares, uses AI and robotic process
decision-making.
3
