Challenges the fintech industry faces with generative AI

Generative AI is shaping every industry, including financial technology.

While generative AI holds much promise for fintech, the industry is still slow to adopt the technology.

One reason for that is the limited regulation that has emerged during the popularization of generative AI. Many financial institutions must be careful in their implementation of the new technology so that they can comply with regulatory requirements.

In this Q&A, Ronen Assia, managing partner at venture group Team8, talks about some of the problems facing the fintech sector in adopting generative AI and how the technology will shape the industry in 2024.

Team8 invests not only in fintech companies, but also in healthcare and other tech- and cybersecurity-oriented enterprises.

Editor's note: This interview has been edited for clarity and conciseness.

How has generative AI affected the fintech industry?

Ronen Assia: The evolution of generative AI has several impacts across different industries in different domains. [In fintech] we're getting more concerns when it comes to fraud, but also a lot more opportunities.

Ronen Assia

When it comes to the application level, a lot of institutions are using GenAI to accelerate or make more efficient direct customer support flows, which is probably the lowest-hanging fruit in terms of what you can do with this technology.

Generally, people are more focused on finding opportunities within the application layer versus finding more opportunities within the back end of the organization. If you think traditionally of a bank, the application is probably 5% of the effort, and the back offices 95% of the effort.

Driving efficiency through this technology makes more sense than just the application layer.

What is one concern for fintech companies in the generative AI market?

Assia: The big question is around regulation. Whenever there is a technology that creates so much buzz and so much concern, there will be regulation in this space -- and I would estimate some very significant regulation.

We see some early signs of new directives, specifically in the European Union, but also President Biden's AI executive order and several others in Asia that are starting to create some noise. But there is no enforcement yet, and there is no clear framework yet.

But we've seen, across the past two decades now, similar cases where regulation really stepped up and became something that everybody needs to comply with. For example, GDPR, the California Consumer Privacy Act and HIPAA.

When you think about GenAI, because of its implications for the organization, it's actually segmented. There's no single point of contact within the organization that just does AI. There's the engineering consequences. There's the data consequences. There's the modeling consequences. There's the application consequences. There's the actual regulation, if it comes to specific verticals. For example, hiring people in New York City now [requires a check for] bias within the models.

So we'll see more specific regulation going toward specific industries -- if it's recruiting people, if it's having models that have to do with credit, with underwriting, with screening clients or onboarding. We can think about different processes within financial organizations that will be probably treated differently.

There's also the more traditional financial-based compliance, accounting practices and things like that, which also, some people say, 'OK, it will make accounting easier or will make legal work easier.' But I don't see a judge yet getting through to the phase of 'Hey, this was done by ChatGPT, and now, please rule in favor or against.' It's much like this debate of if there is an autonomous vehicle, who's in charge if there is an accident?

You can't suddenly hand off all these human-based processes to a machine and say, 'OK, it's done.' So I think that's where a lot of the challenge lies. I do think that in 2024 and 2025, so probably two or three years, we'll see more regulation coming into the space.

What are some of the goals of fintech companies in using generative AI in 2024?

Assia: What I see is companies working on the application level. [For example], trying to replace what has traditionally been the role of the financial adviser -- that's one holy grail. It's almost unachievable today, thinking about replacing a human adviser, because there's so many different aspects that go into the space, but it's definitely a long-term vision, which we'll get there.

Basically, replacing all kinds of services-based financial advice and consulting with machines.

We see a lot of traction using GenAI with taxes, GenAI with legal work, GenAI with compliance, where understanding the text is not just 'write a summary of this Shakespeare play.' It's 'write a summary of the latest legal developments in Iowa when it comes to specific state center DML [direct mortgage loan] regulation.' Today, you would need a fancy law firm, which will cost you hundreds of thousands of dollars to get you that type of opinion. But maybe in the future, this type of work can be done way more efficiently using machines.

When you're dealing with fintech in general, the level of accuracy that is required is so high.

Having said that, the thing that we do see is that you still need the human in the middle, a professional.

If it's in the taxes space, you'd need a CPA who will actually make sure that the machine does what it should, because the technology is simply not there yet. There are concerns as I've stated before about regulation and quality assurance. If you give the GenAI engine a piece of a legal document, it might still be 50% wrong.

When you're dealing with fintech in general, the level of accuracy that is required is so high. The data hygiene factor has to be absolutely spotless if you want to give financial services. It's not just writing a script, birthday card or poem. The level of trust and liability has to be way, way beyond that.

The next sort of jump in technology between 'write me a nice birthday greeting for my 10-year-old son' versus 'write me a good summary of this legal document,' there's a big leap there. With the development of technology, we'll definitely get there, but it will take a bit longer in fintech.

How will the fintech market change in 2024 as it relates to generative AI?

Assia: Every startup I see has 'GenAI' in the first paragraph. A year from now, that will not make sense anymore. It's just going to be that everybody's using it because it's a good and efficient way to use technology. It becomes part of the toolbox. You will not have to explicitly say that you're using GenAI to do this or that.

When interest rates change, it intrinsically changes the product. That's why fintech has been hit harder by interest rate hikes versus other industries. With interest rates stabilizing, I think there will be more optimism within the broader fintech industry. We will see the return of more DTC [Depository Trust Company] plays because the cost of creating an application or cost of creating a business will become lower using some of these technologies.

I'm definitely optimistic. I think we will see more funding, more companies and more exciting stuff that goes to market.

Esther Ajao is a TechTarget Editorial news writer covering artificial intelligence software and systems.