It’s hard to take a nuanced position on AI due to the justifiable anger about it and how it’s being forced, oversold, and abused. But there is legitimate use and technology there. And it’s not just about medical usage, it’s got genuine uses and is genuinely interesting. It’s not “intelligence”. It’s barely even “knowledge” but it’s really quite powerful and useful if you understand what it actually is and how to use it effectively.

I describe it as a bunch of statistical models, collected in various similar but slightly different ways, representing the vast majority of human knowledge ever posted on the internet that is capable of simulating all that knowledge into a surprisingly convincing reproduction of the original information in a wide variety of ways and leveraging a lot of very interesting connections between all that data. Is it perfect? No. Is it accurate? Not in a way that can be relied on. Can it make decisions? No. Is it your friend? No. Is it thinking? No. It’s just pretending, it’s all a simulation, it’s all fake. But the underlying statistical model itself is really interesting and does contain real, valid, useful information.

Having a static, permanent statistical model of the entire internet as it exists at certain points in time, that fits on my phone is wild and if you can’t imagine ways that could potentially be useful not just right now but also in the future you’re not very imaginative.

It’s useful in the same way that demographic data is useful. Demographic data doesn’t let you look into your neighbor’s house and tell you exactly how much money they make, but it can certainly make an educated guess. It might be close, or it might be wildly wrong. That inaccuracy doesn’t make it useless, it’s just useful in a different way because it does contain important, aggregated data that is useful for other things. Just because some people, many people use it wrong doesn’t make it not useful.

The three kinds of lies are: lies, damned lies and statistics. But statistics are still important. You can use them to make wrong decisions and support misleading concepts very easily, but if you’re careful with them and understand their limitations, they can also be very, very powerful. Sometimes too powerful.