giovedì 23 ottobre 2008
lunedì 6 ottobre 2008
Loebner Prize
Can a machine think? It was the 1950 when Alan Turing, in the famous article "Computing Machinery and Intelligence", asked himself (and to the entire world) this burning question. The english mathematician was convinced that it was possible to answer affermatively to the question in the case in which a "machine" (e.g. a computer or a soft-bot) was able to overcome an "imitation game" called Turing test.
Since the 1990 it was created a prize to implement the Turing test in an international competition contest: the Loebner Prize.
At this contest partecipate a set of selectioned "agents" from around the world. This year the competition will be held at the Reading University and there are six finalist systems (Alice, Ultra Hal, Brother Jerome, Elbot, Eugene Goostam and Jabberwacky).
This is the fact. But the critical question is: if a "machine" will be able to fully overcome the Turing test can it be really considered as "intelligent" or, better, as "thinking"? In other words: is it the overcoming of the "imitation game" test a necessary and sufficient condition to state that a machine is able to think?
In don't think so. It's is well known, in fact, (as even pointed out by Marco Varone) that there are a set of "tricks", based on the principle of "dialogue predictability" throught the extraction of keywords from text, that are easily implementable within a computer software but that do not represent a real form of machine textual comprehension.
Since the 1990 it was created a prize to implement the Turing test in an international competition contest: the Loebner Prize.
At this contest partecipate a set of selectioned "agents" from around the world. This year the competition will be held at the Reading University and there are six finalist systems (Alice, Ultra Hal, Brother Jerome, Elbot, Eugene Goostam and Jabberwacky).
This is the fact. But the critical question is: if a "machine" will be able to fully overcome the Turing test can it be really considered as "intelligent" or, better, as "thinking"? In other words: is it the overcoming of the "imitation game" test a necessary and sufficient condition to state that a machine is able to think?
In don't think so. It's is well known, in fact, (as even pointed out by Marco Varone) that there are a set of "tricks", based on the principle of "dialogue predictability" throught the extraction of keywords from text, that are easily implementable within a computer software but that do not represent a real form of machine textual comprehension.
sabato 4 ottobre 2008
Google search 2001. Un salto nel passato.
In occasione delle celebrazioni per il suo decimo compleanno Google si ripropone in versione 2001, permettendo a tutti gli utenti di ricercare informazioni sulla base di quello che era il mondo (anzi il Web) circa 7 anni fa (nonostante, infatti, il gigante di Montain View festeggi il suo decimo compleanno, il sistema di indicizzazione più "datato" che è stato possibile ripristinare è stato quello del 2001, via NY Times). Il link alla versione 2001 di Google lo trovate qui.
Un altro elemento interessante della grande trovata targata Big G è rappresentato, inoltre, dal link "view old version on the internet archive" (che compare accanto a ciascun risultato di ricerca effettuato sul "vecchio" motore) e che permette di visualizzare la "versione 2001" della pagina alla quale si sceglie di accedere.
Un altro elemento interessante della grande trovata targata Big G è rappresentato, inoltre, dal link "view old version on the internet archive" (che compare accanto a ciascun risultato di ricerca effettuato sul "vecchio" motore) e che permette di visualizzare la "versione 2001" della pagina alla quale si sceglie di accedere.
Link: Google lets users search like it's 2001, New York Times.
Iscriviti a:
Post (Atom)
