Would not having a complete foolproof step by step plan to obtaining a nuclear bomb somehow make me wrong then?
The so-called "plan" is simply "fund the R&D, and one of the R&D teams will eventually figure it out, and if not, then, at least some of the resources we poured into it would be reusable elsewhere". Because LLMs are already quite useful - and there's no pathway to getting or utilizing AGI that doesn't involve a lot of compute to throw at the problem.
