Dr. Scott Zoldi, the chief analytics officer at FICO, has been in the industry of building AI models for over 25 years. He warns against getting too comfortable with AI, as many new adopters are using models without understanding how they work. This lack of understanding can lead to biases that may impact customers. To address these concerns, Zoldi recently submitted patents to create an immutable record of AI model development using blockchain technology.
The first patent, which has been granted, aims to provide transparency in AI model development by documenting who performed, tested, and verified the work. This record allows for identifying mistakes, rejections, and remediations, ensuring accountability in the process.
To promote the education and training of future experts in AI, Zoldi sponsored the first FICO Educational Analytics Challenge. This initiative partners with historically Black colleges and universities to train students in working with complex data. The first year’s challenge focused on identifying and mitigating bias in data collected by the Consumer Financial Protection Bureau, with future challenges likely to address payment fraud data.
Zoldi emphasizes the importance of transparency in AI model development and urges new adopters to avoid complacency when using AI. By understanding how AI models work and implementing measures to promote transparency and accountability, organizations can mitigate the risks of biases and errors in AI applications.
To further promote transparency in AI model development, Zoldi’s second pending patent focuses on documenting how the AI model was built. This would enable future monitoring and understanding of the model for years to come.
Overall, Zoldi’s efforts towards promoting transparency and education in AI have been crucial for ensuring that organizations use this powerful technology responsibly.
