Kevin Drum is here to tell you that the AI revolution is just around the corner.
Yeah, well… maybe. The Lake Michigan analogy is great, and it’s a powerful way to demonstrate the power of exponential growth. But the thing is, we haven’t seen a lot of sign that processing power leads to anything that can reasonably be called Artificial Intelligence. The AI field has had a lot of people say, “Well, maybe we can just build a neural net and throw a lot of hardware at it and out will come AI!” So far, they’ve been really badly wrong.
The current iteration of that idea is called Deep Learning, and I don’t have enough AI background to do a thorough job of understanding it. But based on my mere bachelor’s degree in computer science, it’s clear to me that there’s still a leap of faith in saying that we have the method and now all we have to do is let the hardware catch up. Deep Learning does have some new techniques that the last few shots at machine learning lacked, and hardware is better, and who knows, maybe this time there’s a wolf.
Artificial Intelligence is very likely possible. And it could be right around the corner. It could be much less than 20 years out. But anyone who tells you that it’s almost certain to exist by 2040 is selling you a bill of goods.