Is there a limit to how 'smart' AI will get - is the term 'AI' contradictory in its own right ?

There are actually two different types of AI (though as far as I know, the two types have not been labelled differently yet).

ONE - is the one that most of us are used to, the one that makes a computer APPEAR to make intelligent decisions - but in fact the computer is merely following a program. That is: the computer itself is not actually making ANY decision, it is helplessly following program instructions which FORCE the computer to make the decisions that it makes, and the decisions that it has been forced to make have the appearance of intelligent decision-making.

This is the AI typical in computer gaming and other sorts of relatively simplistic interactive computer programs.

I call this "AI where the intelligence is not real intelligence, the intelligence itself is artificial"

Artificial Fake Intelligence.

THE OTHER - is the sort of computers that LEARN, that acquire experiences and make decisions based on those experiences WITHOUT any fixed computer program mandating what decisions the computer makes under what circumstances. There are various types of such processors, one example being a neural net processor.

I call this "AI where the intelligence is real but constructed artificially rather than biologically, the intelligence is real, the components forming the intelligence are artificial".

Artificial Real Intelligence [ARI]

Now, the last I read from a reputable source

- which was probably ten years ago

was that ARI had advanced to the point where the most advanced ARI had mental capacity on par with an ant.

I assume that it has advanced VERY much further since that time

but

realize that "very much further" might be something like

the intelligence of a honey bee

or the intelligence of a dragon fly.

I doubt that it has advanced to the stage

- the intelligence of a mouse

I don't know the current state of what I call ARI

but

there is no reason to believe that ARI has some sort of unsurpassable intelligence limit.

That is: there is no reason to believe that we cannot one day manufacture ARI processors that are more intelligent than human beings.

If we do

and if we set them to the task of designing even MORE intelligent ARI processors

then of course the only limit is the fact that there are limited materials to make a processor.

An ARI that "took over the world" Terminator-style

could feasibly work to create the "ultimate ARI"

that used every resource available on Earth for the construction and operation of the ARI. That would be the limit (unless the ARI employed spacecraft to acquire even more resources), and no telling how smart such an ARI might be.

we don't know if AI is even possible.  its all been in science fiction.  We do call some things ai..but a computer that can recognize a pattern is not a real Ai.  its just trying to imitate one.