I truly wish you are right about all the fear mongering about AI being a scam.
It's something I have been concerned about for decades, similar to the risk of nuclear war or biowarfare. One difference is that nukes and to a lesser extent plagues are more clearly distinguished as weapons of war and generally monopolized by nation-states -- whereas AI is seeing gradual adoption by everyone everywhere (and with a risk unexpected things might happen overnight if a computer network "wakes up" or is otherwise directed by humans to problematical ends). It's kind of like cars -- a generally useful tool -- could be turned turn into nukes overnight by a network software update (which they can't, thankfully). But how do you "pull the plug" on all cars -- especially if a transition from acting as a faithful companion to a "Christine" killer car happens overnight? Or even just all home routers or all networked smartphones get compromised? ISPs could put in place filtering in such cases, but how long could such filters last or be effective if the AI (or malevolent humans) responds?
If you drive a car with high-tech features, you are "trusting AI" in a sense. From 2019 on how AI was then already so much in our lives:
"The 10 Best Examples Of How AI Is Already Used In Our Everyday Life"
https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.forbes.com%2Fsites%2Fb...
A self-aware AI doing nasty stuff is likely more of a mid-to-long-term issue though. The bigger short-term issue is what people using AI do to other people with it (especially for economic disruption and wealth concentration, like Marshall Brain wrote about).
Turning off aspects of a broad network of modern technology have been explored in books like "The Two Faces of Tomorrow" (from 1979 by James P. Hogan). He suggests that turning off a global superintelligence network (a network that most people have come to depend on, and which embodies AI being used to do many tasks) may be a huge challenge (if not an impossible one). He suggested a network can gets smarter over time and unintentionally develop a survival instinct as a natural aspect of it trying to remain operation to do its purported primary function in the face of random power outages (like from lightning strikes).
But even if we wanted to turn off AI, would we? As a (poor) analogy, while there have been brief periods where the global internet supporting the world wide web has been restricted in some specific places, and also there is some selective filtering of the internet in various nations continuously ongoing (usually to give preference to local national web applications), could we be likely to turn off the global internet at this point even if it was proven somehow to greatly produce harms? We are so dependent on the internet for day-to-day commerce as well as, sigh, entertainment (i.e. so much "news") that I can wonder if such is even possible now collectively. The issue there is not technical (yes, IT server farm administrators and individual consumers with home PCs and smartphones could turn off every networked computer in theory) but social (would people do it).
Personally, I see value in many of the points Michael Greer makes in "Retrotopia" (especially about computer security, and also about chosen levels of technology as a form of technological "zoning"):
https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Ftheworthyhouse.com%2F202...
"To maintain autarky, and for practical and philosophical reasons we will turn to in a minute, Lakeland rejects public funding of any technology past 1940, and imposes cultural strictures discouraging much private use of such technology. Even 1940s technology is not necessarily the standard; each county chooses to implement public infrastructure in one of five technological tiers, going back to 1820. The more retro, the lower the taxes. ... This is all an attempt to reify a major focus of Greer, what he calls âoedeliberate technological regression.â His idea is that we should not assume newer is better; we should instead âoemineâ the past for good ideas that are no longer extant, or were never adopted, and resurrect them, because they are cheaper and, in the long run, better than modern alternatives, which are pushed by those who rely on selling us unneeded items with planned obsolescence."
But Greer's novel still seems like a bit of a fantasy to suggest that a big part of the USA would willingly abandon networked computers in the future (even in the face of technological disasters) -- and even if it indeed might produce a better life. There was a Simpson's episode where everyone abandons TV for an afternoon and loves it, and then goes back to watching TV. It's a bit like saying a drug addict would willingly abandon a drug; some do of course, especially if the rest of the life improves in various ways for whatever reasons.
Also some of the benefit in Greer's novel comes from choosing decentralized technologies (whatever the form) in preference to more-easily centralized technologies (which is a concentration-of-wealth point in some ways rather than a strictly technological point). Contrast with the independent high-tech self-maintaining AI cybertanks in the Old Guy Cybertank novels who have built a sort-of freedom-emphasizing yet cooperative democracy (in the absence of humans).
In any case, we are talking about broad social changes with the adoption of AI. There is no single off switch for a network composed of billions of individual computers distributed across the planet -- especially if everyone has networked AI in their cars and smartphones (which is increasingly the case).