Breaking down the state of lending: What’s next for AI and blockchain?

AI is becoming more prevalent in banking, moving from largely internal use cases to external applications. But what are the intricacies of this technology, and what data is required to take full advantage of the capabilities that AI represents? Also, what are the barriers holding financial institutions back from scaling up from initial pilots to strategic roll-outs?

Pablo Padin Fabeiro, Partner, Core Banking & Payment Center of Competence, IBM Consulting, says that first and foremost, AI will help banks personalize products and services for customers through the analysis of large data sources.

“Secondly, AI will help streamline processes - especially in the case of corporate banking - involved in underwriting and on boarding,” says Pablo. “But for this to work as it should you need the right data as well as very good data governance.”

On top of that, Pablo stresses that banks need to train their own small language models in-house so that they are not sharing all of their information to the public cloud.

Overcoming pain points

Ian Kent Harris, Business Development Specialist at Nord/LB, says that most banks recognize that these are steps they need to go through when adopting AI, but that there are still some pain points that they have to overcome.

“These are new frontiers for all of us to navigate, especially US banks,” says Ian. “We can be set in our ways, but we need to take the opportunities presented by AI and blockchain, especially with reduced workforces and when working within strict regulations.”

Ami Ben-David, Founder and CEO of Ownera, agrees that some caution is required. Current applications of AI, such as ChatGPT, are mostly used when people need to access areas where data is openly available: you can ask it anything and it will give you answers based on the public information available out there.

“In banking, you need to combine proprietary data and other relevant data with AI,” he explains. “People should only see the data that they are permitted to see, and that isn’t mixed with other sources that they should not have access to”.

Yet given the progress that ChatGPT has made in the last two years, financial institutions can expect there to be far more complex applications of AI ahead of us, adds Ami.

“We are at the intersection of two industries: one is the digitization and tokenization of assets, the other is AI capabilities. They are colliding - and will create an explosion that will enable a lot of capabilities far beyond what we are seeing today.”

Ami says that this will involve much more than being able to research markets in more detail, for example. “It’s about having agents that can optimize your portfolio in real-time, which is particularly useful if you have a long tail of assets that are all digitized. Suddenly the whole market changes in quite a profound way.”

Ian says that despite those changes, there are still concerns within the banking industry that, in the past few years, data from large learning models has been manipulated and uploaded into AI, creating artificial bias.

Ami asserts that while such problems were common with earlier tools, technology is improving at pace. “I think the problem is coming from a different direction,” he says. “And that is having enough accurate in-house data to avoid either bias or hallucinations.”

Eradicating paperwork

Robert Downs, Global Head of Corporate & Syndicated Lending R&D at Finastra, says that one of the biggest aims for banks from using technologies such as AI and blockchain is to eliminate paper from processes.

Use cases for banks are growing all the time, such as signing transactions with a cryptographic wallet or by typing your name into a digital document, says Ami.

While there is a perfect storm of technologies arriving at the same time, there is also a perfect storm of cost control, resource challenges and regulatory requirements, particularly in Europe, Robert says: “So what are the light bulb moments that need to come on in peoples’ heads for them to realize that these doors are open to them?”

For banks to push these use cases forward and seize the opportunities they create, it will be important for bank leaders to bring everyone in the business along with any process transformation, and not leave anybody by the wayside, answers Ian.

“You need to set clear goals and really inspire people, setting a vision for what you want to do and why” he says. “Leaders need the next talent generation in finance to get on board and help us overcome these hurdles.”

A change in the banking culture could also be required to move from small projects to adopting new technologies at scale, adds Pablo: “Everyone is running pilots and proof of concepts using AI, now we need them to fully take new technologies on board. This is difficult for banks because they are risk control businesses, but they run the risk of being left behind.”

“There are times when the stars align and everything changes fast,” agrees Ami. “All assets are going to be tokenized, including assets and securities - hundreds of trillions of dollars.”

Overcoming technical challenges with blockchains

Technical challenges with public blockchains will be solved, and private blockchains are already inherently more secure, Ami explains. “Private blockchains are in effect just another type of database with different qualities such as mutability. The regulators are beginning to understand there is no difference between putting your ledger on an old-style relational database and putting it on a new style blockchain.”

Blockchain and tokenization can have a perception problem because they are often lumped together with crypto currencies, which are associated with high energy use and gambling. But in reality, they are part of the same technology continuum, Ami explains.

“These technologies learn from each other, but you can’t confuse tokenization with crypto,” he says. “You will have tokenization of real-world assets such as securities, bonds and loans as we go down the track. It will bring dramatic improvements in the way in which cash and payments are handled. We will be able to do instant settlements instead of 35 days, for example.”

It may take time for banks with strong risk controls to completely embrace AI and its intricacies at scale, but with the collision of banking and AI underway, it will be those that take a lead that win the greatest rewards.