Visa hasn't worked for online purchases for me for a few months, seemingly because of a rogue fraud-detection AI their customer service can't override.
Is there any chance that's just a poorly implemented traditional solution rather than feeding all my data into an LLM?
If by "traditional solution" you mean a bunch of data is fed into creating an ML model and then your individual transaction is fed into that, and it spits out a fraud score, then no, they'd not using LLMs, but at this high a level, what's the difference? If their ML model uses a transformers-based architecture vs not, what difference does it make?
Traditional fraud-detection models have quantified type-i/ii error rates, and somebody typically chooses parameters such that those errors are within acceptable bounds. If somebody decided to use a transformers-based architecture in roughly the same setup as before then there would be no issue, but if somebody listened to some exec's hairbrained idea to "let the AI look for fraud" and just came up with a prompt/api wrapping a modern LLM then there would be huge issues.
I run a small online software business and I am continually getting cards refused for blue chip customers (big companies, universities etc). My payment processor (2Checkout/Verifone) say it is 3DS authentication failures and not their fault. The customers tell me that their banks say it isn't the bank's fault. The problem is particularly acute for UK customers. It is costing me sales. It has happened before as well:
I've recently found myself having to pay for a few things online with bitcoin, not because they have anything to do with bitcoin, but because bitcoin payments actually worked and Visa/MC didn't!
For all the talk in the early days of Bitcoin comparing it to Visa and how it couldn't reach the scale of Visa, I never thought it would be that Visa just decided to place itself lower than Bitcoin.
Kind of the same as Windows getting so bad it got worse than Linux, actually...
Visa hasn't worked for online purchases for me for a few months, seemingly because of a rogue fraud-detection AI their customer service can't override.
Is there any chance that's just a poorly implemented traditional solution rather than feeding all my data into an LLM?