DAVOS, Switzerland (Reuters) - Safeguards are needed for using artificial intelligence in financial services to ensure data is accurate and reliable, but only after the opportunities from AI have been identified, London Stock Exchange Group CEO David Schwimmer said on Tuesday.
The European Union has provisionally approved the world's first comprehensive set of rules for AI, with the United States also unveiling an executive order, but Britain has so far held back from bringing in bespoke rules, saying a cocktail of existing rules can be applied for now.
"It's important to have some regulatory guard rails around the use of AI, including verifiability of data," Schwimmer told a panel at the World Economic Forum in Davos, Switzerland.
It is important to avoid incorrect predictions from AI-driven models due to questionable data, he said.
"You have to be careful about putting regulatory restrictions in place before we have figured out what the opportunity is," Schwimmer said.
Finance has long used AI, machine reading and "robot" advice, but Schwimmer said generative AI has the potential to be transformative, with industry participants and regulators trying to catch up with the advances.
Exchanges and other financial sector firms are already heavily regulated, and typically there has often been an adversarial relationship between finance and regulators, Schwimmer said.
A partnership would be better suited for getting to grips with such a rapidly evolving technology, he added.
Charlotte Hogg, CEO of Visa's European operations, said a rush to regulate AI could freeze innovation.
"I don't think we should have the regulatory structure of the grave, but of course, we should have regulatory involvement," Hogg told the Davos panel.