Theorem: “The limit of The Artificial Intelligence”.
The limit of the Artificial Intelligence are not set by the use of machines themselves, and biological systems could be used to reach this goal, but as the Logic that is being used to construct it does not contemplate the concept of time, since it is purely formal logic and metonymic lacks the metaphor, and this is what Gödel’s theorems remark, the final tautology of each construction or metonymic mathematical language, which leads to inconsistencies. The construction of the Artificial Intelligence is an Undecidible Problem .
This consistent logic is completely opposite to the logic that makes inconsistent use of time, inherent of human unconscious, but the use of time is built on the lack, not on positive things, it is based on denials and absences, and this is impossible to reflect on a machine because of the perceived lack of the required self-awareness is acquired with the absence.
The problem of Artificial Intelligence is that we are trying to build an Intelligence system to replace our way of thinking, at least in the information search, but the special nature of human mind is the use of metaphor which lets human beings reach a conclusion, therefore does not exist in the human mind the Halting Problem or stop of calculation.
If you suppose as a theorem, that it is possible to construct a machine, with a Intelligence with capabilities similar to human Intelligence, should we face it as a theorem, we can prove it to be false with a Counter Example, and it is given in the particular case of the Turing machine and “the halting problem” or stop of calculation.
So all efforts faced toward Artificial Intelligence are doomed to failure a priori if the aim is to extend our human way of thinking into machines, they lack the metaphorical speech, because only a mathematical construction, which will always be tautological and metonymic, and lacks the use of metaphor that is what leads to the conclusion or “stop”.
I asked myself the same questions many times, once in a while, in the last two decades, “Is it possible to build a human-like thinking machine? … or it’s a doomed effort?”
The best answer I can give to myself, which is what I now believe, is that it is NOT a doomed effort, and it will definitely happen some day, sooner or later. So, please excuse me for my disagreement.
The Halting Problem might be true for any Turing Machine (TM). And it might comfort us to think that human beings, with our human minds, being able to tell that we are thinking, thus superior to the machine in every way. Since there is no machine can tell (decidable) whether it is running or not.
What I would like to claim is not that a TM-based AI is decidable, but that there will be some (or many – if you will) kinds of AI that can tell if it is running about as good as a human being can.
Here is one possibility I imagined…
Let’s assume that a single-cell organism wouldn’t know if it is dying (since it has no mind). However, a human being can.
In a similar way, a single TM cannot tell whether it is halted yet. But if we hooked 100 billion TM’s together, it could definitely have the same potential as those 100 billion neurons in a human brain to tell whether it is halted yet.
Just wait until we can figure how to compact 100 billion tiny TM’s into 1,700 cubic-cm…