[mmcconnell1618]

Self-learning opens new training opportunities but not at the scale or speed of current training. The world only operates at 1x speed. Today's models have been trained on written and visual content created by billions of humans over thousands of years.

You can only experience the world in one place in real time. Even if you networked a bunch of "experiencers" together to gather real time data from many places at the same time, you would need a way to learn and train on that data in real time that could incorporate all the simultaneous inputs. I don't see that capability happening anytime soon.

[8n4vidtmkvmk]

Why not? Once a computer can learn at 1x speed (say one camera and one mic with which to observe the world), if it can indeed "learn" as fast as a human would, it sounds like all we need to do is throw more hardware at it at that point. And even if we couldn't, it could at least learn around the clock with no sleep. We can give it some specific task to solve and it could work tirelessly for years to solve it. Spin up one of these specialist bots for each tough problem we want solved.. and it'd still be beneficial because they like 10xPhD people without egos to get in the way or children to feed.

Point is, I think self-learning at any speed is huge and as soon as it's achieved, it'll explode quadratically even if the first few years are slow.