00:00Here's everything AI is doing inside your bank in two minutes.
00:05Fraud detection first. Your bank scores every single payment in real time, learning your normal
00:12spending, then flagging anything that breaks the pattern. American Express does this with
00:17deep learning models that score a transaction in under two milliseconds, and it caught about
00:246% more fraud than their old system. Across billions of transactions, that's enormous.
00:31Next, credit and lending. AI can approve a loan in minutes instead of days, using hundreds of
00:39signals instead of one narrow score. Done right, it can responsibly reach people the old system
00:45ignored entirely. Over a billion people worldwide have no traditional credit history at all.
00:52Then there's anti-money laundering. Old systems flagged so much noise that over 90% of alerts
01:00were false alarms, each one costing a human 30 to 45 minutes to clear. AI cuts that false alarm rate
01:09by up to 95% while actually catching more real criminal networks. And customer service. Bank of
01:18America's AI assistant, Erica, has now handled over 3.2 billion conversations, resolving 98% without a
01:28human in under a minute each. But here's the catch. If a model learns from biased history, it can quietly
01:36repeat that bias at massive scale. And the most powerful models can be black boxes. Even the bank can't
01:44always explain why you were denied. That's exactly why regulators stepped in. Europe's AI act now treats
01:52credit scoring as high risk, with fines up to 7% of global turnover. So faster fraud protection, broader
02:00access to credit, real time crime fighting and instant service. But only as trustworthy as the guardrails
02:07behind it. That's AI in banking. Full episode. Link in the bio.
02:12All right.
02:12All right.
02:12care.
02:12Leo.
02:12Leo.
02:12Let usuv
02:1217
Comments