How likely is it that governments would spend up to 50% of GDP or more on a possible AGI if there was a clear path or architecture to achieving it?

And by AGI, I mean actual AGI, not just increasingly capable LLMs. My definition is really ambitious, if humans can do something intellectually, AGI should be able to do it at machine speed and at least at the level of the best humans or organizations. By that definition, I don’t think we’re even close. We don’t have a known architecture for it, we don’t know what breakthroughs are still required, or even whether this kind of AGI is theoretically possible. So claims like “AGI in less than 5 years” seem crazy to me in this subreddit, we will have really strong LLMs but nothing even close to actual AGI. No joke but I feel like we know more about the theoretical possibility of time travel than we do about how to build real AGI. For example, give it one frame of a movie and ask it to make the entire thing: animation, lighting, voices, music, etc., with Pixar level quality. Rendering would obviously still take however long the hardware needs, but the actual work would happen at machine speed. I mean literally giving AGI agents access to PCs and having them do the work the same way humans do today, not generating the movie the way current generative AI creates videos, but actually using the tools, creating the assets, animating, editing, and putting everything together themselves. Ask it to build something comparable to Windows, a modern AAA game, or an entire Moon/Mars landing program, and it should be able to produce the code, designs, simulations, equations, papers, documentation, testing, etc. that currently require thousands of specialists and years of work. The same goes for research: work that takes world class teams years could potentially be compressed into hours or minutes where physical experiments aren’t the bottleneck. Basically, I mean the intellectual work of organizations like Pixar, Microsoft, Rockstar, or NASA compressed into software operating at machine speed across essentially any field. That’s what I mean by AGI, not just a much better chatbot.