The Best AIs are Hiding in Plain Sight
We’ve all been wowed by recent advances in large language model (LLM) AIs, and ChatGPT is making a big impact on pop culture. In fact, this phenomenon has set off a new gold rush in tech, with seemingly everybody trying to figure out how to pivot or integrate “AI” into their plans.
But let’s not forget that AI that is technically more impressive — and definitely economically more impactful — is arleady everywhere, and has been for quite some time.
For example, most white-collar worker might not realize that when they work from home and log into their office systems, their enterprise zero-trust solution feeding numerous signals (source IP address, details of previous logins, time of day, geographic information, etc.) into an AI model to decide whether to allow the login, require additional proof of identity, or flag the login attempt as suspicious.
A conceptually similar application of AI happens every time you use your credit card. You might have noticed that when you travel, large financial institutions no longer require you to “pre-register” your trip to avoid false-positive fraud alerts. That’s because their AI-driven fraud detection is now so good that they can reliably decide whether to require additional confirmation factors whenever (and wherever) you use your card.
The non-tech community may only be slightly more aware that whenever they view a “feed” of news, search results, or e-commerce products on a web site or mobile app, sophisticated AI models are deciding (incredibly quickly!) what to show them personally to maximize their utility, enjoyment, and engagement. And don’t even get me started on how this plays out specifically in online advertising, where the smallest improvements to the underlying AI generate huge profits for the largest ad networks.
I could go on and on with additional examples. But the next time you hear water cooler discussion about the latest splashy applications of AI, keep in mind that the best AIs might be the ones you don’t even see.
