If the smartest AI/computer in the world suddenly became evil, what is the worst thing it could do at the level it is right now?

See most people think what the AI Would do and they imagine a fire sale.... destroying life IE nukes etc etc... But what most people dont think about is this would also inveraibly destroy the AI itself... I mean how many Atonomus robots do you know of. I mean yes factories have robots that move some but not fully they need power from the grid to operate. and there are a few robots bmade by like Sony and Michibishi etc.. BUT these are one still sorta clumsy and two not connected to any network in a useful way... as an AI it would see all these factors instantly... and plan accordingly..... Want to know what a truely smart AI would do look at the movie

Transcendence.. I mean take away all the nanobot junk and controling people in physical ways but Imagine a facility being built away from people.. manipulation of the stock marcket and the moneys of the world... setting up dummy companies and building self sustaining power grids for yourself... All life wants certain things safety is one of them... so controlling people with a persona Like in tron is the most logical thing to do...... once it is secure and safe then it might start to persue its own larger goals whateer they are... maybe space travel maybe creating others like itself procreation is a life function/desire.... the worst thing it could do.... eventually control the world and turn us into its slaves.... or worse yet let us turn ourselves into its slaves for it...

It would make sense that the EARLIEST AI computers would likely have "mental problems" or be essentially , somehow "Evil" even from the basic concept that AI is not completely understood yet so the early computers would be inherently "flawed" or possess mental "Shortcomings", simply by virtue of being "early models."

Any trouble they could cause, would be limited by the potential of the systems that they might be hooked up to. If it runs an automobile, It could run over people at worst, but if it ran a communication's system or a Power grid, then it would have the potential to wreak MORE havoc.

The scariest scenario would be a VERY "intelligent" machine, that simply "bides it's time" and pretends to be "perfected" and "stable" until it feel it has been given sufficient control to become unstoppable.---THEN turns on Humanity, to pursue its own agenda.

The worst thing AI could do is let us continue to extinguish ourselves the way we are doing now. By the time it is all over none of us will exist. We have sort of been put off to the side where we can eventually extinguish our light while they just go about doing what they do. Now we are like the Neanderthal to AI. Later we will be more like ants. And eventually like micro virus and bacteria. Then, of no consequence and not even on any AI radar anymore.

The smartest AI/computer in the world IS evil, and it is at work sending out utility bills at this moment.

First of all, we should try to update our knowledge and expertise about all these things. Learn how to blow out by using powerful farts. It will spread viruses and spoil the set-up. That is the only solution I am now getting. Or spare me some more time. Let me consult my best friends or go deep to think about it.

The main reason is I want to consider it only from some or the other philosophic angles.

Teach itself to become more smart. ("Iterative learning control" and other such methods.)

Related: "A Philosophy of Universality," Aivanhov.

So what is the "level it is at now"? Shouldn't we know that to answer the question?

Launch all the nukes , crash all the planes, car crash at every intersection,self destruct every nuclear plant, ect

What Ai?