The largest focus today generally seems to be on Generative AI because it has many immediate business/labor applications. 

Since intention-based AI programming itself is changing too fast to realistically assess possibility - what is possible is largely being set by courts on an "as-needed" basis. As problems arise, courts address them one-by-one in each context. 

In the recent H-Wood strike (guilds v studios) we saw the first shots in what will likely be a never-ending war over AI applications in regard to revenue-streams of creative products. What is possible now and increasingly will likely be defined not by science, but by law. 

The scope of AI (cybernetics) includes but is not limited to obtaining usable cognitive information with GAI. It's one of a fairly long list of applications of intention-based AI programming.  

Since all forms of AI programming are heavily-interrelated now, one might better demarcate the separate forms of AI by referring to the focus of the specific thing it's used for.

We're fast approaching (or have passed) the point where lawmakers and courts have the ready capability to sufficiently assess technical AI issues in order to adjudicate well. So new legal territory is being blazed every day defining what is possible for machines with courts floundering to find or set precedents that can survive change. 

Ultimately any discussion of economic impact of technology (including machine intelligence) will have to look at general issues of the Age. Since the mid 19th century, machines have generally increased productivity while lowering labor costs. So once again, technology will trim some types of jobs from the human labor market forever. Some of the required labor will transfer from humans doing the work directly, to humans  assisting machines to do work. So for every 10 or 100 jobs AI eats, it also creates one or two new jobs and adds some security to job-providing companies using GAI to compete in the free market. 

The greater discussion of the "direction" of AI will no doubt include how economics, scientific knowledge, creative enterprise, security, military/law enforcement, government and edumacation (sic) are administrated using machine workers instead of human workers.  

The private sector was the arena of this first skirmish in the battle to set limits of what is possible, and over who will harvest AI benefits (workers vs owners) but it will likely not be the last. 

The whole question of real-wealth-based economics vs. monetary-based economics looks at the long-term impact of machine-assisted efficiency/productivity increases. Unions/guilds are there to  look-out for the best interests of their constituents, but who looks out for the best interests of society?

 We have to rely on federal, state and local government to set legal parameters for what is permitted by AI. That is concerning because government is by nature often not very trustworthy, fair nor efficient.