Parisi: « Urgent a plan to attract American researchers to Europe »

«The Italian government plan to attract American researchers who want to leave Trump’s America is a right thing. As Europe we have a duty to welcome them but also the opportunity to grow our human capital exactly as happened in the 1930s When many scientists had to leave Europe due to racial laws and went to America where they contributed to the scientific and technological growth that the United States had. Obviously the situation has been different since then, but the US government threats to the independence of universities are real and many researchers are pushing the problem of where to continue their work. The plan presented by Minister Bernini It is an important signal but the dimensions are not sufficient ». The Nobel Prize for Physics Giorgio Parisi76 years old, is in a lounge of the Accademia dei Lincei where a meeting of the Board of Directors has recently ended which for the umpteenth time has made an appeal to increase the funds for research.

Another Nobel Prize for Physics, Enrico Fermi, who in 1938, due to the racial laws, left Italy to land first in New York and then in Chicago where he had a key role in the « Manhattan Project », which led to the realization of the atomic bomb, was also part of the Linxes. In 1932 it was Albert Einstein who left Nazi Germany to get marked to Princeton (the university that was seen to suspend 210 million dollars of funds from the American government); And after him many others left. They could do so because in 1933 a fund was created to welcome scholars fleeing Nazism: « The Emergency Committee in Aid of Displaced Foreign Scholars ». Parisi says: «We need a European scale plan. We have a great opportunity to recover the delay in many sectors, in particular on artificial intelligence « . The meeting had been set to talk about the latest developments in artificial intelligence and « Disclaimer», The tour in the universities that Corriere della Sera and Cineca have announced to make everyone more aware of this umpteenth technological revolution. Salvifica or even outgoing, as some say, for the human species? «I think it will be a happy ending fairy tale. Obviously there may be problems but it is up to us to prevent there.  »

The end of the world was feared each technological revolution. It is not that by dint of shouting « to the wolf to the wolf » when the wolf will really come, we will not notice it anymore? Are we underestimating the risks?
« It is clear that if an increase in productivity is not accompanied by a reduction in working hours, unemployment is created, we will have to supervise this. Another problem will be the relationship with written texts and I see an impact on the world of information: if people are satisfied with the summaries made by the artificial intelligence of the news published in the newspapers, who will subscribe to the newspapers more? But if the newspapers close, what will artificial intelligence summarize? ».

When the generative artificial intelligence came, someone called it « a stochastic parrot » who guesses a succession of words. Is it becoming something more?
« It has been trained like a parrot in Stocastic, there is no doubt about this, but by force of guessing it is becoming a parrot who understands something. »

Does it start to have an idea of ​​the world?
« I would say no. Artificial intelligence is absolutely disengaged: it can speak of sadness, however, it has never tried it and has no idea what it really is. Practically connects words with other words without then having the meaning of what happens. The other limit is the fact of having hallucinations, that is, of giving false answers with absolute certainty ».

Once she said she had convinced the artificial intelligence that 4 by 5 ago 25. Today chat gpt knows that 4 by 5 does not do 25.
«The models have improved enormously and a whole series of errors no longer make them. But every now and then they invent answers: recently a chatbot invented quotations for my article. The key question to be done is always: are you really sure? Once he told me: no, I pulled to guess ».

Professor in Cristianini says that the latest models are in terms of human intelligence. Do you agree with who says that this is the last year in which the human species is the most intelligent on planet earth?
«But no, because in the meantime I would like to see these models that invent new things. Artificial intelligence cannot invent the future. He has the knowledge of everything that has written humanity and recombine it. Our creativity also puts together everything we have read, learned, heard, proven and lived and from there generates ideas. But the future does not create itself by pulling to guess but trying to combine the experience with an idea of ​​the world. That the chatbots don’t have ».

So do I agree with the philosopher Miguel Benasayag who claims that « chat gpt doesn’t think »?
«We should define exactly what thought is and suppose that humans think. It is clear that the machines do something different even if, as Hamlet says, there is a method in their madness. But they do completely consistent things and in some cases they help us a lot. For example, the translations have become very good, sometimes better than those made by a human being. I used it to check the quality of the translation of my book in Chinese and I discovered the inaccuracies of human translators ».

How does the work of you scientists change? Is it helpful to you?
«One thing impressed me very much: I had written by hand formulas in my bad calligraphy and artificial intelligence not only did he read it, put them in order in a mathematical format and added the right comments between one formula and another. Just a good parrot! ».

Almost all this game is played between Silicon Valley and China. The marginal role of public actors is a non -trivial problem, right?
« Absolutely. There is a need for a transnational initiative such as CERN, or a large laboratory in which there are a thousand scientists who work together. To the point where we are the fundamental thing is not so much hardware, that is, investments in supercomputer, because those are there ».

And in Italy there is Leonardo, one of the best in the world as a calculation capacity.
«The fundamental point is to put the scientists in contact, make them work in the same place, make them discuss. To get a supercomputer, just detach a check, instead having brains is much more difficult; For this reason it would be essential to start from a European artificial intelligence center physically located somewhere and divert a small part of European funds on human capital ».

Professor, she has been saying this for a few months. What can we do to make you pass from a wish for a project?
«I have been talking about it for eight years, I only added to the last. We have lost a huge amount of time and in the meantime artificial intelligence has evolved. Now we have to avoid resignation, to think that we are now back and there is nothing to do. The game is still open and Europe can play a role ».

And Italy? On physics we have a formidable tradition but would you recommend to a researcher on Ai to stay in Italy today?
«The problem is that Italy does not invest in research. Staying in Italy can be convenient in some niches but for the rest the lack of programming and funds does terrible damage ».

In conclusion, what is your « disclaimer » on the current revolution?
« We have all the skills to be able to develop artificial intelligence but we must give resources to research otherwise the best young people go away. »