Uncertainty

take by me on some random subject

So there are two kinds of AIs as I see it: 1. real AIs – those that we actually have, you write to them and they respond to you, nothing fancy, nothing truly intelligent; 2. fictional AIs – these are human-like intelligences, can think the same way as humans, plus optionally do heavy calculations thanks to the computers that they run on.

Now, I want to add another fictional type from the future, sort of combined AI. Key difference is it would act like a real human-like or general AI, but at its core it wouldn't be much different from what we have in the real world, the people using it just won't be able to break it or hit the limit as easily as with the current AI. So in a nutshell this would be AI that we, the users, cannot disprove that it's a general AI. Which makes it different from 2nd type from above, because people believe that 2nd type is true AI, on the contrary this 3rd type people would utilise as a true AI, but won't treat it as real intelligence, despite it being very hard to tell the difference.

This is a rough idea, if anyone but me reads it. Don't expect some genius thing from it, it's only an attempt to somewhat tune down too-good-to-be-true AI helpers in fiction that I randomly thought about.