The technological singularity. What happens to our world when AI can do a thousand years worth of intellectual work over the weekend?

[Anomalism]

20 years?

AI didn't exist ten years ago.

In 10 years we went from glorified adding machines to abstract problem solving.

We're on the verge of understanding the relationship between stochasticity and periodicity. Once we have that, warp drive will be very close.

Let's do it. Lol

I'm ready.

 

[scruffy]

Let's do it. Lol

I'm ready.

Yeah, you and I are. But the kids... how can I say this... we're reaching the cusp of complexity, where the things we'll be asked to do are complex.

An example: a simple political poll. Used to be, you go to statistics school, then you become a pollster. You learn, how many people you need to ask, you learn about bias, stuff like that. Same thing in music, you just ask people who they like, do the statistics, and then you get "greatest rock band of all time" or some such nonsense. "Hillary will win".

But these days... it's all about social media, right? Who's your circle of friends, who did you friend or unfriend, and if you unfriended someone was it because of politics or because of Lynyrd Skynyrd? So the number of factors in your polling increases, and with AI the number of factors grows exponentially.

So kids, no matter how smart they are, reach a point where the complexity becomes overwhelming. And that's when, the reason we had "disciplines" becomes apparent. Kids today, they're being taught to "ask AI". Instead of learning the basics, they look for the canned computer library that gives them the answer they want. They don't need to know "how", they just need the answer.

I've been watching a great series of stats vids by an anthropologist, he sees this same problem. He's working on a "workflow" to guide students through the probabilistic aspects of AI. The thing is, he never talks about AI. But at the end of his course, you know exactly how AI works, and how to make use of it.

Here's an example, he's talking about "generative models". Which is an AI term, right? Nope - it's a statistical term that originated in statistical analysis more than 100 years ago.

 

[scruffy]

Oh - DAG is directed acyclic graph.

 

[Votto]

Let's do it. Lol

I'm ready.

Ok, but first, how about a nice game of chess?

 

[Anomalism]

But the kids

Many of you would probably think of me as a kid.