Intelligence everywhere: How AI is transforming the future of lending

Innovation is key, said Benoit Lafort, Regional Vice President at Ncino. “We want to understand where you see AI as a component of the lending process,” he told a Business Reporter Breakfast Briefing at Amsterdam’s Amstel hotel.

Current use

 

AI chatbots such as ChatGPT , or in-house built models, are already making tasks such as powerpoint presentations and writing pitches much more efficient. But more complex tasks still often require a human to check output. 

 

Current use in the lending sector by banks includes using AI to analyse financial statements. An online bank said that in 50% of cases, where templates are used, a human is no longer needed to check decisions. However, if statements come from accountants, with different fonts and formats, the model still has difficulties. One bank has started testing using AI to process diligence reports, but is struggling with the parameters.

 

Only one large bank has incorporated AI into both input, decision making and output, allowing an in-house built AI to decide on cases at a corporate level, similar to what is done in retail. “In retail we have models. The challenge is to move that towards mid-corp and large corp and to weed out easy decisions,” one executive said. This requires a leadership that is on board, and years of testing to ensure the model works, as well as ensuring proper documentation to obtain regulatory approval.

 

Opportunities

 

What banks care most about is whether AI can help with their P&L. Ideally AI helps processes become more efficient and decisions more accurate. For some banks, it’s also about impact.

Two large fields of opportunities for AI were identified. The first is in optimizing process, ensuring loans or guarantees can be issued faster. Already, the timeline for decision making has reduced drastically, if the relevant data is available.

 

On the input side of things, automatization and AI can help free up talent to spend more time on client facing interactions. AI is also already being built for the output side, to help justify decisions.  As to making the decision themselves, executives had different opinions on the capability of current AI models and the need for human oversight. Future uses that were envisioned included using AI to actually define parameters. And it would also be desirable if AI can help identify clients that don’t meet current lending criteria, but still would be good candidates for a loan.

 

The second opportunity is in terms of advising clients, where AI could be used to identify opportunities and develop business propositions. In an ideal world this would be some kind of all-knowing chatbox which can analyse qualitative data coming from diligence reports, external data sourced from the internet, reports, public reports, transaction data and not only summarise but identify what isn’t clear yet, to for example develop strategies of how markets develop.

 

 Concerns

 

Concerns about data quality and data protection were paramount. The main concerns of banks is how outside parties use their data, Tim Mussche of Ncino said. Smaller banks in particular have concerns about the availability of data they can feed AI models. The quality of data is also a concern, with the profilation of desinformation and fake news. How do you ensure the quality of the inputs?

 

Ethical concerns raised included how to avoid bias in a model.  Sustainability was also raised, as the data centers used to power AI use huge amounts of energy.

 

Customers might also have privacy concerns. “Do our clients want us to have so much information about themselves?” one executive asked. “I don’t.”

 Concerns about regulatory approval were also questioned. “Shouldn’t we be more disruptive towards regulators?” one executive from a large bank asked. “We need to prove to them that based on what we or third parties have created, or what we’ve used; the services we provide, that our analysis is better.”

 

Business case

 

While many executives recognised that there was a lot of excitement within their banks for AI, the hype can create unrealistic expectations. “There is a gap between the story in the media, perception and where we are,” said one online bank executive.

 

Currently the business case for investing in AI is not always there if you compare the costs needed for implementations versus the benefits that can currently be obtained, several executives said.  Cooperation might be a solution to this, and there was enthuiasm among participants to meet again, also perhaps with regulators.

 

Some banks are further along in their use of technology, including AI, than others. Especially for big commercial banks the use of AI offers opportunities due to the large amount of available data from other sectors, including retail.

Boards see the advantages of AI, how it can help drive business, create more efficiency and upsell things. But questions remain as to how AI can best be used, what the risks are and how to control the use of AI.

 

---

 

To learn more, please visit: www.ncino.com