Business

The Shift Toward AI-Driven Credit Scoring: Innovation and Oversight

The reliance on traditional credit scores to determine an individual’s eligibility for mortgages, loans, and mobile contracts is facing increasing scrutiny, as industry experts and consumers grapple with the accuracy and fairness of existing systems.

For many, credit scores can be a source of sudden instability. Sophie Hope, a Brighton-based nail technician, discovered her credit score had dropped to ‘very poor’ due to identity fraud. Unbeknownst to her, accounts had been opened in her name and debt had accumulated over several years, only surfacing when she applied for a mortgage. Her experience highlights a recurring criticism: traditional credit scoring is inherently ‘backward looking.’

image

Funmi Olufunwa, a consumer finance lawyer and founder of Hoops Finance, suggests the country is ‘overly obsessed’ with these scores. Traditional models, utilized by the three major credit agencies—Experian, Equifax, and TransUnion—often fail to capture a full picture of affordability. Furthermore, they frequently disadvantage ‘thin-file applicants,’ such as young people or recent immigrants, who have not yet established a formal credit history.

In response to these limitations, some firms are turning to artificial intelligence to analyze broader datasets. Companies like Upstart and Zest AI are marketing machine-learning models that process thousands of variables to assess creditworthiness. Upstart states its model can increase lending accessibility across demographic groups. Mortgage expert Sarah Tucker, CEO of The Mortgage Mum, notes that AI has the potential to help lenders identify patterns and create more individual assessments of risk, which could improve financial inclusion. However, Tucker emphasizes that consumers need transparency regarding the information used in these decisions and the ability to challenge inaccuracies.

image

These technological advancements have prompted calls for caution. Sandra Wachter, a professor of Technology and Regulation at the University of Oxford, warns that without proper safeguards, AI risks replicating and entrenching past social inequalities. ‘Not putting any safeguards in place will fortify, foster, and enhance discrimination,’ Wachter stated. While companies like Upstart maintain they are actively investing in automated, objective evaluations to identify and mitigate bias during model training, critics argue that automation must be paired with strict accountability.

Some industry participants are seeking alternatives that avoid AI entirely. Evangeline Atkinson, who founded the credit reference agency Noggin HQ, was motivated by her own struggle to obtain a phone contract despite being financially active. Her company, which received FCA authorization in June, analyzes current account transactional data to help lenders interpret spending habits without using AI, which Atkinson argues risks inbuilt bias and the improper inference of protected characteristics.

image

Meanwhile, established credit agencies are increasingly integrating AI into their operations. Experian announced a partnership with OpenAI in August, allowing users to access credit scores via ChatGPT. Equifax has launched a platform named Equifax Amplify AI for credit assessment and fraud detection, and TransUnion began integrating Google’s Gemini models into its analytical assets in March.

As the industry evolves, the adoption of these new methods may depend on generational attitudes. Olufunwa notes that while younger, tech-savvy generations may feel more comfortable with their data being utilized in this manner, it remains to be seen if older demographics will share that sentiment as firms move away from traditional scoring methods.

Mia Reynolds

Mia Reynolds covers companies, entrepreneurship, leadership, and changing industries. She turns complex business developments into accessible stories, helping readers follow corporate decisions, market shifts, and trends influencing the global business landscape.
Back to top button