GenAI has already trained on the entirety of human content, and it still screws up basic inquiries. I think it will get worse as it cannibalizes its own hallucinated data and the problem of hallucinations gets worse as a result.

The US will not regulate, but regulations originating from international jurisdictions will roll downhill to some extent and force the bad actors in the US to establish some guard rails in order to continue to be able to offer their products in these big international markets.

I think the most significant change in GenAI will be social. People’s poverty and isolation is only getting worse, and it’s likely there will be an economic crash due to the concentration of wealth in this sector and its failure to deliver on its promises; However, there is genuine demand for frictionless relationships and that is one thing generative AI does very, very well. In a decade these models will have artificial bodies and people will be literally publicly dating their AI companions. These companions will also be another great way for companies and governments to spy on people, so expect them to get more cost-effective in order to ensnare as many users as possible in the coming years.