A study conducted by researchers at Lehigh University revealed concerning racial bias in loan recommendations made by chatbots for mortgage applications. The experiment, which included 6,000 sample loan applications, found that Black applicants were more likely to be denied loans and offered higher interest rates compared to identical white counterparts. This bias was particularly evident for applicants with low credit scores, with white applicants being approved 95% of the time while Black applicants were approved less than 80% of the time.
The study highlighted how AI algorithms and large language models are being used in industries like finance, healthcare, education, and the judicial system. While these technologies have the potential to streamline processes and reduce costs, they also have the potential to perpetuate bias and discrimination. Understanding that flawed training data and historically biased information can impact outcomes, researchers emphasized the need for regular audits to refine AI tools and prevent bias.
The findings of the study underscored the importance of human involvement in using decision-making AI tools fairly. Legislators in four states have introduced laws aimed at preventing algorithmic discrimination, and there is a growing call for government regulation to ensure the ethical and unbiased use of AI technologies. Researchers stress the need for industry-wide standards and frameworks to validate and certify AI models, ultimately improving trustworthiness and reducing the potential for discrimination. As these technologies become more ubiquitous, it is crucial to ensure that they are deployed in a fair and equitable manner.
Source
Photo credit tennesseelookout.com
