I'm with the bet-the-house guy. I've worked for huge organizations since 1990 and I really think that even a mid-grade student project could make decisions as well as the management I've known, which includes multiple Fortune 50 CEOs. When I say "known" I mean "known" and not "been ten layers under."
What I think will happen is CEOs will be glorified actors that are tasked with delivering the decisions some ML system has made. It'll be like the "white guy in a suit" role that some Chinese companies have.
There are strong reasons to believe that tasks that are new to humans and are considered difficult, will be easy to AI. Like programming, finance, strategy, etc
Whereas things humans have developed over millions of years and think are easy, like vision, dexterity, etc might be quite hard and the last things to be automated.
I feel like a lot of comments about what ML will do start with a misunderstanding of what ML actually is.
Saying ML will make a CEO’s decisions is like saying Excel will make CEO’s decisions. Neither is anything more than a system for keeping track of correlations.
ML models don't make decisions, but there certainly are ML systems, within which are ML models embedded with data flows, business logic, etc., which make "decisions" or "adjudications" or "classifications" or whatever term you are trying to avoid here.
Neither do most CEOs. What they do is best guess which fully evaluated option for (d)investment they like better based on gut feel. Most business-oriented decisions are basically stack ranking the options presented.
What I think will happen is CEOs will be glorified actors that are tasked with delivering the decisions some ML system has made. It'll be like the "white guy in a suit" role that some Chinese companies have.