Artificial Intelligence in Banking: between Technology and Ethics
The financial sector is one of the most delicate and studied fields for the employment of Artificial Intelligence in recent years. From the use of conversational chatbots for assistance to trading, algorithms allow banks to reduce errors and significantly improve the performance of data processing and computation.
In particular, according to a McKinsey analysis, AI is mainly used in robotic process automation for operational tasks, virtual assistants for customer service, and machine learning techniques for fraud detection and risk management.
However, allowing algorithms to have a major role in financial and economic decision making processes also raises some ethically significant issues: is it correct to delegate decisions that so delicately affect people’s lives to a computer?
=====
Do you want to stay up to date on the world of Artificial Intelligence? On November 29, 2021, the workshop AIABI 2021 will take place, where companies and the academic world will discuss the applications of Artificial Intelligence in business: join us too!
=====
Lending DocAI: Artificial Intelligence for Finance
One of the most relevant and controversial applications of machine learning in the banking world is its use in mortgage lending. For example, Lending DocAI is the Artificial Intelligence system produced by Google that promises to “automate the processing of documents related to mortgages”: through financial algorithms, the software evaluates an individual’s eligibility for a mortgage. In this way, it minimizes risks for the bank and makes the analysis process more efficient.
By analyzing data related to “tracking people’s financial habits, such as the punctuality in paying bills, loans, or signed contracts, as well as consumption habits and preferences, often through the applicant’s social media,” a score, called credit score, is calculated and used in predicting the risk of default.
In this way, Lending DocAI risks making discriminatory choices against the weaker social classes, who are more prone to payment delays. Moreover, the limited amount of financial data makes it more difficult to evaluate individuals from these social classes, putting them at greater risk of being rejected by the bank’s Artificial Intelligence.
Martha Bennett, Principal Analyst at Forrester Research, argues that computers “can crunch vast amounts of numbers, applying different algorithms. They don’t make mistakes, unless they’re badly programmed.” Perhaps it is precisely this rigidity that is the true limitation of algorithms. The adoption of Explainable AI methodologies, which guarantee individuals the so-called “right to explanation” regarding decisions that affect them, becomes not only a technical challenge but also, and above all, an ethical duty of Data Science.
For more information: