Embed Notice

@leigh my bet is that ML will get miniaturized / optimized and their feedback loop will start to become near-instant, and after that we'll have dozens of little ML programs all over the place, but they'll mostly be in toolkits or invisibly doing stuff.

In other words it'll change a LOT but the ML programs won't be the feature, and eventually the net result will be kind of like computer animation: sold as a way to reduce workforce, but actually increases it a lot.