Note on AI
Much is said about the potential of AI, both to replace human tasks and to surpass them, but little is said about how AI can be many things but intelligence, perhaps artificial, but never intelligence.
As has been discussed since the beginning of AI in the 1960s, the name AI was abandoned for a less charming and less pretentious name, which is why early AI researchers intended to simulate the human brain, hence the names used from the beginning such as:
a) electronic brain;
b) Artificial Intelligence;
c) neural networks;
d) expert systems.
Since the beginning of the Aristotelian era, human knowledge has been officially divided and defined into very distinct and discrete areas, thus the sciences and human intellectual activities emerged, which were divided into:
a) Philosophy;
b) Medicine;
c) Arts;
d) War;
e) Religion;
f) Politics;
g) Astrology.
Religion ceased to have the exclusive answers to all universal questions and had to give way to other areas of knowledge that strongly challenged the primacy and totality represented by sacred books and beliefs, superstitions, fetishism, and animism that were responsible for explaining the origin of the universe, life, luck, the future, and the destiny of everything and everyone, explaining the laws of nature and the laws of causality and the control of the universe, the organization of things, animals, and events of all kinds.
Then, in the modern era, Positivism emerged, where the controversial Auguste Comte summarized the knowledge from Aristotle and Galileo on a new scientific methodology, which is always evolving, to the refutationism of Karl Popper, in short:
a) scientific methodology;
b) organization of the scientific research method;
c) Presentation of contrary evidence;
d) The impermanence of provisional knowledge.
So, the decisive division of human knowledge into specific areas is the first and definitive evolution of science when it separated itself from Philosophy, Mysticism, Religion, tradition, and superstition.
In the 1950s, there were few engineering chairs:
a) military;
b) naval;
c) civil;
d) electrical;
e) agronomic;
f) mechanical;
Needless to say, electrical engineering chairs were subdivided into other specialties:
a) electronics;
b) power;
c) microelectronics;
d) communication;
e) digital;
f) networks;
g) mathematics and digital Boolean operations;
Therefore, it would be natural for AI to specialize in its specific branches of competence, since AI could not deal with each and every area of human knowledge; thus, the name AI was replaced by Expert Computing Systems, or Expert Systems - ES. This new name never caught on because it lacks marketing appeal, doesn't sell well, doesn't excite, and doesn't create expectations of complete brain development.
Divisions of AI, or Expert Systems (ES)
1 – Analog;
2 – Digital.
The analog ES branch is known in the market as commercial AI that manipulates texts, capable of making interpretive heuristic inferences about a subject within the norms of grammatical textual interpretation of the language, with excellent translators, seeking the standard form of grammar, punctuation, accentuation, pronoun placement, separation, and periods according to the linguistic norm of each language in syntactic analysis.
Its algorithm is based on rules of semantics and statistics on a broad database from a set of stored discourses separated by subject, specialty of human knowledge in the humanities, such as:
a) History;
b) Politics;
c) Religion;
d) Ethics;
e) Laws;
f) Norms;
g) Geography;
f) Arts;
g) Biographies;
h) Biology;
i) Geology;
j) Astronomy;
k) Pedagogy;
And so on, with more than 2000 areas of expertise to form your frame of reference for developing responses, such as ChatGPT, and others of the kind.
2 – Digital SE
Digital systems are based on knowledge from the areas of mathematical, or exact, sciences, such as:
a) Mathematics;
b) Physics;
c) Chemistry;
d) Statistics;
e) Engineering;
f) Geometry;
h) Astrophysics.
Therefore, the work is easier while being limited by the cognitive limits of Mathematics and Physics themselves.
Processes in SE based on Mathematics have severe limitations that are inherent to Mathematics, for example:
a) Matrices up to level three;
b) Algebra works with up to three independent variables simultaneously;
c) Integrals only up to the third degree; d) The limits of physics that works with the exclusion of variables through modeling and abstraction;
e) The loss of information in derivatives with equations containing constants;
f) Margin of error in limit and trigonometry calculations;
g) Various universal constants such as pi, 0!, division by zero, exponent zero, division of primes among themselves, fractions, square root of two, and many other ad hoc definitions;
Therefore, the conclusion reached is that the limits of AI or SE are set with the same limits of human knowledge, which cannot be violated in these algorithms.
AI or SE elements, and moreover, the fields of knowledge of AI are quite segmented and cannot be randomly combined without review and supervision of human knowledge intelligence through human intervention, because they are based on human knowledge, rules, guidelines and limits of human knowledge, and cannot transcend or surpass them.
Nenhum comentário:
Postar um comentário