It's Saturday, June 7, and keyboards have caught fire in the Royal Society of London, with volunteers exchanging text messages with unknown users. With their questions they seem as if they want to meet their interlocutors - e.g. their name, age, musical tastes or hobbies.
Στην πραγματικότητα, όμως, προσπαθούν να καταλάβουν αν εκείνος που απαντά είναι άνθρωπος ή κάποιο chatterbot, δηλαδή ένα «διαλογικό» πρόγραμμα τεχνητής νοημοσύνης (Artificial Intelligence-A.I.). Στο πλαίσιο του πειράματος που διοργανώθηκε από το πανεπιστήμιο του Reading, πέντε chatterbot υποβλήθηκαν στο «τεστ Turing», δοκιμασία που επινόησε ο Βρετανός μαθηματικός Alan Turing το 1950, προκειμένου να διαπιστωθεί αν μια μηχανή είναι το ίδιο ευφυής με τον άνθρωπο.
As predicted by the test, this would be the case for any program that managed to "trick" a sufficient number of volunteers into the five-minute dialogues they had with them. The Russian software "Eugene Goostman" succeeded in 33% of cases, presenting a 13-old boy from Ukraine. The performance of "Eugene Goostman" is a record and for the first time exceeds the limit of 30% that the mathematician has set, making Reading's University declare the first software that passed the "Turing test". The news made the round of the world as a step closer to the era when machines would have been "gifted" with intelligence like ours. However, some wondered if this era would prove to be disastrous for humans, reminding HAL 9000 from the "Space Odyssey".
From the Matrix to Hawking
Many media recalled the "Space Odyssey" a few days ago, along with "The Matrix", "The Exterminator" and "I, the Robot". A common denominator of films is that they describe a future era in which "thinking" machines turn against their creator. A scenario that came to light on the occasion of Stephen Hawking's statements that "the full development of artificial intelligence may mean the end of the human species."
Οι δηλώσεις έγιναν στο BBC στις αρχές Δεκεμβρίου, με την ευκαιρία καινούργιου software συνθετικής ομιλίας που ανέπτυξε η Intel γι' αυτόν, το οποίο «μαθαίνει» πώς εκφράζεται ο διάσημος φυσικός, για να του υποδείξει ποιες λέξεις θα ήθελε ενδεχομένως να χρησιμοποιήσει και, αν πάρει από τον ίδιο το «πράσινο φως», να τις εκφωνήσει. Ο Hawking δεν αρνείται πως τέτοιες «πρωτόγονες» εφαρμογές τεχνητής νοημοσύνης έχουν αποδειχθεί χρήσιμες, υποστηρίζοντας όμως ότι ο άνθρωπος θα κινδυνεύσει στο μέλλον αν αναπτύξει μια «σκεπτόμενη» μηχανή. «Θα μπορούσε να αυτοβελτιώνεται με γεωμετρική πρόοδο, ενώ το είδος μας εξελίσσεται πολύ αργά», είπε.
Hawking also agrees with Elon Musk, the man behind the first automotive electric car industry. Tesla and spaceX company SpaceX, who has warned of "the greatest possible threat to our existence". "It's like agreeing with the devil," he said at a MIT conference last October, and eight months earlier he had invested $ 40 million in the Vicarious FPC, together with Mark Zuckerberg and Ashton Kutcher, to " I keep track of developments, "as he had said on CNBC.
In 30, 40 or 60 years?
Both Musk and Hawking refer to what will happen soon after scientists develop the first machine with "strong" artificial intelligence, meaning as intelligent as a human. This landmark since the 50s has been called the "technological singularity", and one of the best descriptions of it has been given by Ray Kurzweil, analyst, inventor and director of the Engineering Department of Google.
For example, in his book "The Singularity Is Near" (2005), from that moment on, the machine will begin to improve without any interference, resulting in its intelligence overtaking the whole of mankind. However, although Kurzweil expresses reservations about whether algorithms can be developed so that such machines have a moral code, they are not worried they will want to enslave our species.
But when does artificial intelligence reach the human? Google's manager predicts that 2045 will happen, while other scientists will be in 60 years later than today. In his book "Superintelligence: Paths, Dangers, Strategies," Nick Bostrom, director of the Oxford University Institute for the Future of Humanity, quotes the results of 4 polls, to which 170 scientists were asked when it is likely that machines will acquire such intelligence, to take up almost all human occupations. Predictions reached 50% for 2040 and ejected to 90% for 2075.
"They will lack self-knowledge"
There is no shortage of experts who have opposing opinions - even for "Eugene Goostman" scientists have claimed that the digital persona distorted the rules of the test. The same applies to the provisions for "technological uniqueness", which e.g. Professor Tony Cohn from the University of Leeds, speaking to the BBC, moved on a century later.
On the contrary, other scientists say it will never happen. "What is always missing from machines is consciousness, that is, the ability of man to understand himself. Besides, to encode human intelligence algorithms, we should know exactly how it works, which I think is impossible to do, "Dr. Dimitrios Kallis, president of the Hellenic Society of Artificial Intelligence and assistant professor at Open University.
Dr. Ioannis Vlachavas, professor of Informatics at AUTH, agrees that consciousness will always be what distinguishes man from his creations. "Even if a software evolves to answer the most complex questions, it will never acquire self-awareness, that is, it will never "know how it knows", like a human. Also, the initiative will always remain a human one feature, in the sense that a machine will never manage to do something different from what it is programmed to do," he says to "K".
Training… programs
All researchers agree that artificial intelligence has exploded in recent years. From a technological point of view, an important reason is new approaches that make programming more flexible. "Essentially, these are 'machine learning' techniques, that is, algorithms that show in a software how it will infer the patterns that are repeated in the huge amount of data with which we feed it," explains Mr. Vlachavas. So, for example, Google "trained" in 2002 a program to recognize cats, after having "seen" 10 million snapshots with the nice quadrupeds on YouTube. In the same way, IBM's Watson supercomputer managed to beat the two best players in the Jeopardy video game a year earlier.
To the new approaches, one has to add the huge funds that technology giants are investing in artificial intelligence - "the" case "of an expert became about the same as a rising rugby star", Business Week wrote in 2013, commenting on " transfers "of scientists from Silicon Valley companies. Companies that already have I.X. without a driver, even applications that "borrow" Watson… IQ, helping doctors make diagnoses.
At the same time, however, a recent UN summit issued a resolution warning of the dangers of the spread of "intelligent" weapon systems that can identify targets and destroy them without human intervention. "In addition to intelligence, all applications that come will reflect the intentions of those who have developed them, which in some cases will be at least controversial. In that sense, I think the conversation that Hawking has just started is just beginning, "concludes Mr. Kallis.
Landmarks in Artificial Intelligence
1950: In his article "Computing Machines and Intelligence" the English mathematician Alan Turning formulates a thought experiment that a machine would have to "pass" to prove that it is truly intelligent. The "Turning test", as it was called, is still used today for this purpose.
1956: The first conference on artificial intelligence is being held in Darmouth, which also introduced the term.
1963: Entitled "Computers and Thought" and edited by Edward Feigenbaum and Julian Feldman, the first collection of articles on artificial intelligence is published.
1968: MIT is developing a chess software, MacHack, the first to ever take part in a tournament.
1979: As part of his PhD at Stanford University, Hans Moravec developed the Stanford Cart into the first autonomous vehicle. The vehicle managed to cross a room full of chairs, avoiding obstacles without outside intervention. It took 5 hours.
1997: IBM's Deep Blue supercomputer defeats Garry Kasparov, world champion in chess.
2005: A stand-alone vehicle developed by Stanford University defeats the second Grand Prix of the American Darpa, driving without 211 driver miles into the Mojave desert.
2011: IBM's Watson computer defeats the two best players in the American Jeopardy knowledge game.
2011: Apple launches iPhone 4S with Siri, the first "smart assistant" on a mobile phone.
2014: The software "Eugene Goostman" is proclaimed by the University of Reading the first software in history to pass the "Turing test".
Source: kathimerini.gr
