Artificial intelligence (AI) has and is continuing to transform the fintech industry as we know it and it’s looking unlikely that its impact will diminish any time soon. For example, a recent study revealed that 70 per cent of global fintechs are using AI today, and the technology is predicted to dominate the market by 2025, with over two-thirds of fintechs believing AI is the technology that will have the biggest impact on the sector over the next five years.

This comes as no surprise given its ability to quickly analyse massive quantities of data to derive important insights and information, resulting in it being used by thousands of businesses on a global scale to create efficiencies and recognise patterns that can significantly improve decision-making.

However, this also creates challenges — some of the world’s greatest entrepreneurs and thinkers from Bill Gates to Elon Musk have predicted that AI will “make jobs irrelevant”. In his book, End Times , Bryan Walsh argued that just as humans have used intelligence to rise to the top of the food chain despite being physically weaker and slower than many animals, so too could machines.

However, when it comes to AI redefining the ‘human touch’ in the fintech industry and more specifically the commercial lending and credit analysis space in which we operate, we’re firm believers that its future lies in enhancing humans, not replacing them. There is not enough data to produce and fit a general model that would accurately assess all corporate credit analysis cases. To perform this task in a fully manual fashion requires the credit analyst to perform a very large number of tasks — some of which can be automated given machine learning techniques applied to the data we do have.

This hybrid approach is a pragmatic compromise where computers perform various tasks to allow the credit analyst to be more efficient, but the analyst remains in the driving seat and is able to train the models and direct and shape the final outputs to ensure they are coherent and understandable. This also means that it is not necessary to solve the entire suite of problems before automation is of some help to overall efficiency, with analysts plugging gaps in the process that are not yet automated.

Not for full automation

In fact, due to the complexity of the space, we at OakNorth don’t believe full automation is a desirable end goal, and aim instead to achieve 80 per cent automation, with a human analyst always involved in the process. This critically allows human judgment to always have an influence on the outcome and helps ensure understandability of outputs.

This way of working — believing machine should be enhancing the human, not replacing the human — is crucial to our operations as we never make a credit decision based on the data alone. For example, as part of OakNorth Bank’s direct lending approach in the UK, we invite prospective borrowers’ management team to our credit committee where they can discuss their borrowing needs directly with the decision-makers. This meeting coupled with the data on the business helps the bank to make a more informed credit decision. If the data doesn’t seem to meet the bank’s rigorous lending criteria, or the management team don’t seem to have a clear vision or business plan, we always give feedback, and there is the possibility to apply again in the future once the issues have been addressed.

The key to successfully navigating commercial lending and credit analysis lies in leveraging AI to minimise time, effort and risk at the back-end, so that relationship managers have more time at the front-end to build stronger relationships with prospective borrowers, build a granular understanding of their business, and offer a delightful experience.

The writer is Co-Head of India, OakNorth