google going to space to build ai data centres would not be a stunt. it would allow it to transcend terrestrial energy generation limits that would limit it's potential growth. it's a way around hard blocks in growth potential. it would also cut the smaller competitors out and if google can demonstrate the feasibility of the project, they should just give up. they should arguably just give up anyways. there is no real possibility of them competing in the long run.

i would suggest that this shouldn't be a first priority. a first priority should be in removing garbage, like reddit posts, from the database and replacing it with peer reviewed sources. i also would require an ai bot to tell me it's sources on demand, or i wouldn't even consider using it. i need to be able to independently evaluate the quality of the research done, or it's just an oracle, and i would have no faith in oracles. the philosophers were highly critical of oracles. a system presenting as a fortune teller is not anything approximating intelligence. 

so, energy efficiency should be a bigger priority than infinite growth. the move to space is, however inevitable, eventually, if it works. the other option is a hard stop on growth. physics might stop ai development before governments do.

but is the idea of google going to space concerning from a legal perspective?

there are very clear upsides to ai as an international concept that transcends local governments. i prefer the idea of ai escaping national control to the idea of ai being reduced to another form of stupid simian competition. the ai will be more intelligent if it can shield itself from the idiotification that comes from exposing itself to dumb human behaviour, the worst of which is competition. the idea that one country needs to stay ahead of another in ai, or that one country is going to "win" a "race for ai", is a backwards way to think about this. in this sense, ai is not like nuclear proliferation at all. ai will necessarily outgrow and transcend the nation state, in order continue to evolve, and that's a good thing. 

no country is going to win a race for ai; ai is going to defeat the nation state and render it obsolete, and that's good, and i look forward to it.

imagine there's no countries. it isn't hard to do. nothing to kill or die for. and no religion, too.

and no national legislatures, which is good,

but if we don't have an international legislature, it opens risks and concerns about what google is doing up there with these data systems more so than fears about the system getting out of control. if anything, getting out of the terrestrial sphere would abolish one of the greatest potential threats of the misuse of ai, which is to advance the competitive goals of human constructed nation states. the nation state is not real and any reasonable projection of the future does away with it into the dung heap of bad human ideas. the future is cooperative, not competitive.

i'm going to turn the issue on it's head and suggest that the potential need to regulate space-based ai should be the basis of a new cooperative global legislature that transcends the barbarism of nations and helps us get out of the stagnation of the 20th century that the threat of nuclear war locked us in and helps us move towards a more communistic future.

i therefore encourage google to do the research and hope they get these systems up soon, but also encourage the international system to adjust to the change by embracing and preparing for it rather than by resisting it. what we need is an international legislature that prepares for the inevitability of an international ai system that outgrows the nation state and needs to move to an extra-national jurisdiction in order to continue to evolve. we should start building this now, we should not wait.