history_01.md 3.9 KB
#History of Artificial Intelligence
Definition
Artificial Intelligence is a discipline with some goals, such as:
- solving complex problems And these goals are reached by artificial agents that think as humans
There is a lack of unique and universally accepted definition.
A proposed definition is in the following schema
| Think as humans | Think rationally |
|---|---|
| Act as humans | Act rationally |
History
General Problem Solver (or GPS) is a program developed by A. Newell and H. Simon and it is an attempt of reproducing the human ratinal process. Alan Turing wrote a paper "computing machinery and intelligence" that focused not only on intellectual processes but mainly in acting processes. Turing Test is based on evaluating a machine only looking at its behaviour and not its internals. Aristotle defined human beings as rational animals.
We as human being are constantly feeling the gap between intentions and actions and is related to free will (libero arbitrio)
Acting rationally
Is explaines by means of a rational agent
Artificial Intelligence
Is a young and constantly varying discipline, for example in the 1990's making chess playing programs was at the edge of the discipline while now the actual goals are like making a robot move in an unknown space.
AI is also between Science and Engineering while the second part is prevalent today, the first part is the one that give birth to it.
The origin
The origin can be reconducted to the tradition of humans trying to create similar of themselves.
- Heron of Alexandria Was the first creator of Semiauthomatic machines called autòmatha (Che destano meraviglia) And he was both a matemathician and an engineer
- Ramon Lull: was a philosopher who wanted to build a machine that is able to reproduce the human process of tought, he called this machine "Ars inveniendi veritatem" and with this he tried to prove that there is only one truth and this truth is the cristian one.
The important idea that Lull introduces is the necessity of representing our knowledge in a symbolic way and deriving knowledge without altering the initial concepts (like modern first order logic)
Lull built this machine and went to north africa to convert muslims using this machine but he got killed there.
Leibniz also tried to create a formal system to obtain truth with a tool (he would have used is to win discussions agains Newton)
Babbage
Boole is the creator of the agebrization of logic, inventing the propositional logic
Frege invented the first order logic while the notation we use today is from Peano
The precursors
The precursors comes both from 1940 and are the following:
Computer Engineering The first computers were built like Z3 or Eniac.
Cybernetics Was a discipline devoted to communication and control of regulatory feedbacks both in living being and machines.
The official birth of Artificial Intelligence was at a workshop in summer 1956 and the object was in ten years to build a machine that can simulate human learning or intelligence.
In 1969 the neural networks model was heavily criticized by Minsky and as a result had a stop in research
In 1966 during the cold war the fundings were lessened and there was a crisis of artificial intelligence The only funded project by the government was machine translation but lacking notions of linguistics, they build only a semantic translator.
After 1969 Expert systems were introduced, and were meant as systems supporting human experts in a particular field. During this years Language translators were improves as semantic+syntactic.
From 1980 to today AI vecame an industry and the biggest tech companies became interested.
In 1986 neural networks were rediscovered, in particular back-propagation was reinvented by four different research groups. This is a type of sub-symbolic approach
The last step introduces the idea of intelligent agents and particularly on groups of agents.