In many fields of R&D, the effort required to maintain a steady pace of improvement grows exponentially. Maybe you develop AGI one year by spending $X, and the

In many fields of R&D, the effort required to maintain a steady pace of improvement grows exponentially. Maybe you develop AGI one year by spending $X, and the next year you need $1.2X to make it 110% as good. For a while, you can keep investing more, and the productivity improvements from AGI also help a bit. But eventually the pace slows down, as the world economy is too small to sustain it. Maybe self-improving AGI just the next technological advance required to sustain 2% annual economic growth.