JPM exploring synthetic data for AML, software engineering

JPMorgan is exploring the use of synthetic data to streamline its anti-money laundering and software engineering processes. Synthetic data refers to datasets that have been stripped of personal identifying information; developers can run simulations and test models on these datasets while avoiding biases and protecting original data, Deepak Paramanand , head of synthetic data product platform at JPMorgan, told Bank Automation News at FinovateEurope in London this week. [caption id="attachment_185980" align="alignright" width="300"] Courtesy/Bloomberg [/caption] Synthetic data operates like a sandbox for large language models, Paramanand said, adding that it can be used for testing large language and AI models. “For example, for every million transactions, you have only 10 money laundering or fraud transactions,” so the sample size to improve the AML [anti-money laundering] model is too small, he said. “The synthetic data can help you create 700, 7,000 or even 70,000 synthetic money laundering transactions” to test your model’s accuracy. By using the technology, JPMorgan has found that it can improve AML model accuracy by 15%, Paramanand said. For software engineers testing their large language models, “the whole software development life cycle around discovering the problem, designing the problem, developing the software, you need data,” Paramanand said. Developers can introduce different variables to make software more robust in testing, which can “achieve about 25% more productivity,” he said. Synthetic data can help make AI models more accurate before they are rolled out in the real world, reducing the number of glitches and inaccuracies, Paramanand said. Synthetic data will overshadow the use of real data for AI model development and deployment by 2030, according to a June 2022 report by think tank Gartner . Other financial institutions, including BBVA , Financial Conduct Authority of the United Kingdom and American Express , are joining JPM in using synthetic data for different use cases. The $772 billion BBVA is using synthetic data in its AI factory to replace real data while testing artificial intelligence models to protect consumer privacy, according to the bank’s website. Join us for Bank Automation Summit U.S. 2024 in Nashville, Tenn., on March 18-19! Discover the latest advancements in AI and automation in banking. Register now.

JPM exploring synthetic data for AML, software engineering

Synthetic data helps boots AML model accuracy by 15%

Citizens receives patent for its Digital Butler

Podcast: Citizens’ approach to new tech

Related Posts

TD Bank Head of Analytics, Intelligence, AI Ted Paris to speak at FinAi Banking Summit

Eltropy AI Voice Agents integrates with 13 banking systems, including Fiserv

First Citizens Bank Head of AI Ash Kaduskar to join FinAi Banking Summit as fireside speaker

Podcast: Citizens' approach to new tech

Stay Informed with Our Newsletters

EMERGING FINTECH DIRECTORY

The Buzz Podcast

SPONSORED

Hiding in Plain Sight: How to Use Data to Spot Consumer Accounts Being Used by Small Businesses

Is Your Cash-Handling Infrastructure Ready for the New U.S. Banknotes?

Just Released! 2025 Strategy Benchmark

Connect

Welcome Back!

Retrieve your password