A recent report from a financial technology expert at the Bank of England has raised a red flag against the extensive use of AI in various industries, including finance. Although the inception of AI promises better efficiency, risks loom large on financial institutions that embrace AI.
The report, published by Kathleen Blake, explores the problem of AI bias and its potential consequences. It also urges the industry to address the pressing issue.
The report categorizes AI bias into two distinct groups. These are biases originating from the underlying training data and biases stemming from the results produced by AI models.
Eventually, these issues reflect the inherent biases of their creators and developers.
Blake further notes that it’s impossible to mitigate the bias based on the model’s output, given that bias rooted in the training data is far more complex.
Removing data points associated with characteristics like gender or ethnicity often turns out to be ineffective. Blake also compared this data bias to the historical practice of redlining in mortgage lending.
He explains, “In other words, the remaining, non-protected features could act as proxies for protected characteristics”.
Societal Bias and the Role of Data
Societal bias, as described in the report, stems when norms and negative legacies from society unintentionally seep into AI systems.
For instance, Blake cited the Amazon recruitment algorithm recommending more male candidates over female candidates due to historical data showing that males were traditionally hired more often.
This example shows how AI systems tend to retain existing biases.
The report also emphasizes that the financial sector comes under threat from AI bias. For instance, AI models charge higher interest rates to Black and Latin mortgage customers compared to their white counterparts.
This fuels financial inequality, leading to a compromise in trust in the banking mechanism.
Moreover, the ‘black box’ nature of proprietary AI models continues to be a concern. Many financial institutions depend on these models for different purposes. Their collective actions may have a significant impact on the economy.
Given that these models are opaque, the outcomes tend to be unpredictable.
While any catastrophic event is yet to occur, there’s a chance of this destabilization becoming real, as in the 2021 case where Apple and Goldman Sachs were investigated for offering smaller lines of credit to women based on algorithmic decision-making.
The report from the Bank of England highlights the urgent need to address the AI bias issue in the financial sector. While the potential of AI is immense, its adversaries in fueling biases cannot be overlooked.
The consequences of unchecked AI bias, as explained in the report, can lead to instability in financial institutions and the broader industry.
Therefore, it’s imperative to prioritize fairness and transparency while deploying AI to address these risks and ensure an equitable financial system.
>>> Read full article>>>
Copyright for syndicated content belongs to the linked Source : TechReport – https://techreport.com/news/bank-of-england-warns-of-ai-bias-threat-to-financial-sector/
