Also not an expert, but from what I can tell, the difference is, as you say, processing power. But also the instant accessibility of huge gobs of data which can

Also not an expert, but from what I can tell, the difference is, as you say, processing power. But also the instant accessibility of huge gobs of data which can be used to train the models on which these ML systems rely. So it's a perfect storm of personal computing power which anyone can pick up for almost pennies, and an Internet which gives access to data. Put the two together at enough scale and you've got magic. The real kicker though, the one that changes everything, is the introduction of open source. By moving it all outside the walled gardens, suddenly anyone on the planet with a PC and a bit of skill can start tinkering with this stuff to do 'what ifs' and launch new concepts. It's totally wild! :)