DR. IAIN BROWN
GENERATIVE AI
This data drought is particularly acute in regulated sectors. Healthcare providers sit on vast troves of patient information that could revolutionise medical AI – electronic health records, diagnostic images, treatment outcomes – but GDPR and similar regulations make sharing nearly impossible. A pharmaceutical company developing a rare disease treatment might have access to only dozens of patient records globally, far too few to train robust AI models. Financial institutions face similar constraints with customer data, whilst government agencies struggle to collaborate across departmental boundaries, even when working toward common goals.
The result is what researchers call‘ data islands’: isolated repositories of valuable information that remain largely untapped for AI development.
DR. IAIN BROWN
TITLE: HEAD OF DATA SCIENCE NORTHERN EUROPE
COMPANY: SAS
INDUSTRY: TECHNOLOGY / ANALYTICS
PhD-qualified data science leader with decade-plus experience spanning banking, AI, and machine learning. Author of“ Mastering Marketing Data Science” and Adjunct Professor at University of Southampton.
How synthetic data can improve AI performance SAS’ s research provides compelling evidence that synthetic data can actually improve AI performance. Working with a major UK financial services provider on personal loan assessments, his team encountered a classic machine learning challenge: severe data imbalance.
The bank’ s historical data contained thousands of examples of customers who repaid their loans successfully, but relatively few examples of defaults. This imbalance is typical in financial services – most customers are good customers – but it creates a training problem. AI models become excellent at identifying
96 December 2025