What i find interesting is the other article of geohot regarding AGI.

It seems that its financial possible for a handful of companies to learn everything.

It doesn't matter how we solve 'work':

It can be AGI, it can also be the already existing massive global scale Reinforcement Loop we all feed through using ChatGPT and co, it could be to compute RL or by buying experts teaching this knowledge to some AI system.

Companies also start to put the 'human' part into the agentic layer.

A while back anyone was somehow a benefit even if they did some kind of shitty work. Today i don't think this is true anymore. I would prefer to manage some avg ai than a shitty person.

This will and is already disrupting human lives.