Italy's Intesa embraces AI to tame bank supervision "information overload"
Technology
Lisa reads texts much faster than humans and can identify correlations that are not obvious
MILAN (Reuters) - Intesa Sanpaolo has introduced an artificial intelligence tool it has designed to wade through thousands of publications on banking supervision, Italy's biggest bank said on Tuesday.
The machine learning tool, dubbed Lisa or Linguistic Intelligence for Supervisory Awareness, uses language processing algorithms to scan documents for correlations and patterns of meaning to help predict future trends.
Lisa reads texts much faster than humans and can identify correlations that are not immediately obvious, Intesa said, citing as an example the connection that wasn't evident until a few years ago between climate change and credit risks.
Intesa has a dedicated team of people who worked with Lisa, validating its results but also expanding its awareness of banking regulatory issues.
"Banking supervision is an area where it is fair to speak of information overload, with truly massive and exponential content production," said Walter Chiaradonna, head of Intesa's supervisory strategic steering department.
Chiaradonna listed "practices, interviews, statements, texts and in-depth studies that in turn generate a proliferation of information that is unmanageable without adequate support."
Intesa said the European Central Bank also used artificial intelligence tools (SupTech) to help sift information such as the minutes of boards meetings or thousands of datasets and documents required from the more than 100 larger banks the ECB supervises directly.
