On October 10, Fintech Nexus hosted a webinar sponsored by AI-powered credit risk descisioning platform Provenir.
Joined by Provenir, OakNorth, iwoca, and LendInvest leaders, Peter Renton discussed the outlook for lending in Europe for 2024 and how AI has shaped the landscape.
AI’s Big Appeal
The past year has seen AI developments take the world by storm. In finance, an area in which the technology is considered to make a significant difference is lending. Enhancing efficiencies in credit descisioning and the ability to process larger datasets quickly, among other applications, AI and automation, is set to birth a new era of digital financing.
![Valentina Kristensen, Director of Growth and Communications at OakNorth](https://www.fintechnexus.com/wp-content/uploads/2023/10/Valentina-Kristensen-Director-of-Growth-and-Communications-at-OakNorth.jpeg)
Of course, AI is not a new technology for the lending space. “We’ve been using AI and ML models since pretty much since day one,” said Valentina Kristensen, Director of Growth and Communications at OakNorth. “(It’s powered) predictive analytics in terms of our scenario analysis, for example… And in portfolio risk management, regulatory compliance, and in the fabrication of our first ever TCFD report.”
The developments made in machine learning have allowed lenders to process large datasets that learn from ongoing usage. This has led to improvements in credit descisioning and assessing risk.
“The area that we’re using AI and machine learning models most is loan portfolio management,” said Rod Lockhart, CEO of LendInvest. “So during the lifetime of a mortgage, as we collect ongoing data related to that mortgage, how’s the likely behavior of that mortgage changed over time. We then use that to focus more or less on a specific loan.”
![Rod Lockhart, CEO of LendInvest](https://www.fintechnexus.com/wp-content/uploads/2023/10/Rod-Lockhart-CEO-of-LendInvest-218x300.jpeg)
He explained that LendInvest then planned to evolve the model to use the data analysis for decisioning, an area others have had success in applying AI.
“Where we really apply artificial intelligence is in our lending and scoring algorithms,” said Christoph Rieche, Co-Founder and CEO of iwoca. “So self-learning, based on information that we receive from our customers, are they repaying or not repaying. That is automatically fed into recalibrations of our model as and when the model detects signals that warrant change in respect of weighting. That is happening in a fairly continuous way.”
AI’s Energy Consumption and ESG Alignment
However, the use of AI in lending can come at an environmental price.
In a recent analysis published by Digiconomist founder Alex de Vries, it was estimated that In a middle-ground scenario, by 2027, A.I. servers could use between 85 to 134 terawatt hours (Twh) annually, a similar energy usage to entire countries. While his analysis focused on the AI industry as a whole, the technology’s increased application to financial products could make a significant impact on their carbon emissions.
This can be a challenge for lenders that aim to align their ESG practices with global net zero objectives. Two of the lending representatives spoke during the webinar about their environmental objectives for the future.
Kristensen spoke of OakNorth’s strategy, stating that their objective was to reach net zero on Scope one, two and three emissions by 2035. She also described their strategy for supporting the ecological practices of their clients and their transition to reach net zero.
RELATED: Fintech’s Scope Three Opportunity
To an audience question of how lenders can fit AI’s high energy consumption within their ESG goals, Kristensen responded, “It’s like anything with the journey to net zero…short term, there may be a spike in the carbon footprint of businesses in certain areas if they’re deploying AI in a material way, but then hopefully, you’ll see those gains as things progress.”
“It’s a very difficult balance because I don’t think you’d be able to achieve the long-term goals without seeing some of those short-term spikes.”
In his analysis, de Vries stated that there was a possibility for AI to reduce energy requirements as the technology matures. Enhancing the efficiency of AI could lead to lower quality hardware requirements and the level of energy needed for their use.
The Challenge Of Data Sources
At the core of the AI boom is the abundance of data. As the world becomes more digital, the financial system has a wealth of new data sources to pull on for lending.
![Louis Garner, VP of Client Success in EMEA for Provenir](https://www.fintechnexus.com/wp-content/uploads/2023/10/Screenshot-2023-10-11-at-17.06.54-232x300.png)
“When it comes to new data sources, I think the world is your oyster,” said Louis Garner, VP of Client Success in EMEA for Provenir. “When it comes to data, I think one of the challenges that everyone faces is how we operationalize it all in a fair and consistent and compliant manner.”
He explained that for many organizations, the amount of data available can be overwhelming. While AI can help process the information, questions are continually posed about the correct use of data sources and when they should be implemented.
“I think a lot of organizations are still getting to grips with how they use what is currently at their disposal today,” he continued. “I think we will see the alternative financial data around utility data or rental data play more of a key part (in the lending process). Then, we may move into some of the social and behavioral data. But that leads back to how we use it in a fair, compliant, consistent manner.”
Lockhart agreed and explained that for property lending, a huge number of data points are collected, which can create challenges.
“In the property space, there’s still a long way to go before we get to fully automated lending in the more specialist areas. We’re collecting around 100,000 data points on each loan,” he said. “The big challenge is making sure that we’re processing and focusing on the useful data we’re collecting.”
“We’re analyzing a tonne of these data points through APIs, but ultimately trying to present a case in a way to allow a human to make an ultimate lending decision.”
Customer Demand For Automation
Automation of lending decisions has become a goal for many lenders in the space. While full automation may still be a long way off for some areas, such as specialist property loans, consumers’ demand has increased.
“Consumers are driving the need for these automated decisions, in my opinion,” said Garner. “We essentially set out from ‘how do we save money as organizations by straight-through processing’, but in terms of the expectations on consumers now, I think that automated decision is no longer reserved for those lower ticket items. It’s an expectation that we can service that.”
He explained that lending organizations are increasingly looking for methods to fulfill that customer need, and Provenir had positioned itself to assist.
“I think from a technology perspective, (Provenir has) got a role to play in how we allow organizations to make quicker, faster straight-through decisioning to satisfy those needs. In terms of the role of Provenir, we look to provide a platform that gives people the ability to build those decision flows, act upon them, and change them quickly.”
![Christoph Rieche, Co-Founder and CEO of Iwoca.](https://www.fintechnexus.com/wp-content/uploads/2023/10/Christoph-Rieche-Co-Founder-and-CEO-of-Iwoca.--243x300.jpeg)
Other representatives in the webinar explained that while some automation had been achieved, at times within seconds, more complex and large loans still required human input.
“In our world, an automated decision is the one that someone can draw down funds from without any further intervention,” said Rieche. “It’s all fully automated 100%. That works really well for some of our customers.”
“Where we’re doing enormous work is where we are phasing in manual intervention or manual intelligence, in a completely unbiased way. No decision is actually taken by humans, but in some of the decisions that the system has taken, there is information that is provided by a human because it’s just too complex for the system to read. A lot of our work revolves around how that information is fed into the system, without any human bias.”
Responsible Lending Focus Going into 2024
Turning towards the future, webinar participants focused on the trends they would be watching going forward into 2024.
For Garner, and Provenir, the attention lies on the developments of Consumer Duty for the lending space and how that will feed into responsible lending. Garner explained that with embedded finance, access to credit had become more freely available, making a focus on responsible lending critical.
“We’re seeing, from an integrated customer journey, the quickness in which you can check out now and choose that embedded finance option,” he said. “It’s worlds apart from where it was maybe two, three years ago. So we’re certainly seeing that seamless integration and that interconnected customer journey being driven through that demand for customer experience.”
“We’re certainly looking at the role technology can play in ensuring we are lending responsibly…particularly around embedded finance.”