"Why would the amount of research being done stay the same, if the amount of money coming in goes up by a factor of 10?"

Number of publications. And lack of other strong Bayesian evidence. Money does not correlate well with thinking capacity; if you dump $20 million into a startup, its intelligent output will (on average) drop off rapidly.

"I guess they might spend it on advocacy, or buying hardware, or something, but surely what it would take for your comment about utility to be correct would be for them to do *nothing* with it. Why would they do that?"

I have no idea what SIAI's current budget is, or how they spend their money. I'm analyzing it using black-box efficiency, how much goes in versus how much comes out.