VISA

Company launches AI-based payment system for cards

Visa said on Wednesday that it has developed an artificial intelligence system capable of automatically approving or rejecting credit and debit card transactions. Remember: Visa is one of the largest payment card networks in the world, judging by the number of transactions.
The decision to approve or reject transactions is usually made by the bank that issued the card. But banking networks can fail due to natural disasters, faulty software or other reasons. Visa said that as of October, its autonomous and autonomous system, capable of self-checking transactions, will be available to banks that subscribe to the service. "This technology is an important first step to help reduce the impact of disruptions," said Rajat Taneja, president of Visa Technology.
The new service reflects the growing use of AI in the banking sector. According to research firm International Data, banks will spend 1 billion in 2020 on artificial intelligence technologies, and by 2024, annual spending will grow to 4.5 billion on initiatives such as fraud analysis and investigation.
The new Smarter Stand-In Processing (SSTIP) service uses deep learning and was created by Visa data scientists and software engineers: the company holds three patents for this technology, two of which are pending. According to Mr. Taneji, several disruptions in the banking networks affect several million credit and debit card transactions each year, often leading to unwarranted transaction rejections. In that case, cardholders sometimes need to call the bank for help. Stores and banks may lose revenue if transactions are not completed due to failures in the interbank network.
"These problems affect business, but our main objective was to make the consumer environment more friendly," said Tanezha.
The service will be triggered automatically when the Visa network detects that there is no access to the infrastructure of a specific bank. According to Carolina Barcenas, senior vice president and chief of Visa Research, the older version of STIP, which used a simpler rules-based model, performed worse and was not very accurate.
SSTIP does not use hard and fast rules - the new deep learning model is more advanced, trained to filter billions of cardholder activity data points to discover relationships on its own. For example, it can automatically know that a specific cardholder's transaction was normal and must be approved based on the user's past payment and transfer history, as well as the merchant's location in relation to the cardholder, the time of purchases and so on.
"The model reveals many of these relationships and creates the kind of intelligence that analysts have had to create in the past. The new version of the technology works better and more accurately than the previous one," said Ms. Barsenas. In testing, the model deep AI learning accounted for 95% of the bank's decisions to approve or reject transfers, she said. According to Visa, this technology has more than doubled accuracy compared to the old method. Both versions will continue to coexist, but the more advanced version will be available to customers as an additional service.

AVnews