Pascale Fung is a software engineer, a lecturer in two departments of the Hong Kong University of Science and Technology and also the director of CaiRe, Hkust’s AI Research Center. When we met her, on the occasion of her participation in SAIConference, the event on artificial intelligence organized by the SAI Hub of Sienaexplained to us that “we promote collaboration between East and West in the field of AI, trying to encourage the exchange and circulation of information” between the two parts of the world.
However, it is not to make us tell you this that we wanted to talk to her: we did it because she is one donnabecause it is of origin asian and why it is a source authoritative in an industry that tends to be dominated by white American males. Which is already a problem now and will risk getting bigger if not addressed.
Generative AI interview, 5 questions (plus one) to Hiroshi Ishiguro by Emanuele Capone 18 March 2023 Pascale Fung during her visit to Siena
3 real ways AI will help us
“It’s true: this is a decidedly more masculine world, but in less than 10 years it will change, to the point that already half of my students are women – Ed told us at the beginning of our chat – it is important that it changes especially in the field of generative AI, which must give correct answers and cannot be developed only by those with traditional skills in mathematics and statistics”. The reason why it is important that it changes is also the reason why it will change: “About twenty years ago it was very rare to have women who approached the so-called stem materialbut with AI it is easier because they touch us and will touch us in many aspects of our lives and therefore to develop them we don’t just need the usual professionalism typical of the hi-tech world“.
In China, by the way, the current discrepancy is less true even if it seems equally true: “In my field, there are more women than in the West but they have very little visibility – Fung explained to us – It happens because AI is above all a product, a business that borders on the business world. Which in turn is still very masculine.”
Whether all this is resolved or not, what is certain is that artificial intelligences will have a significant impact on the lives of all men and women. “They will be and already are useful above all in the medical field”, Professor Fung told us, giving us a couple of concrete examples: “Both in Asia and in Europe, the populations are getting very old, we are living longer but we also get sick more, we will need AI and robots that take care of us and also that they work for us”. And then? “Besides, AIs are already very good at detect diseases from premonitory signals invisible to human doctors (such as some forms of cancer, ed) and will also be fundamental for speed up progress (which is the same thing they say in IITed) and to resolve political issues and complex problems, such as those related to climate change”.
“They’re just machines”
AI will help humans solve human-made problems, but that’s not why Fung seemed to have any little or no fear, contrary to many male colleagues. Males and Westerners, which according to her is an important detail: “This fear of AI, of the end of humanity that they could cause and in general of the end of the world, is typical of the West – she told us, reasoning out loud – The scenario of armageddon and the biblical apocalypse is precisely typical of the Bible and Judeo-Christian culture”. And it is “something that even unconsciously influences us since we were children, as I well know because I grew up and went to school in the West as well. In China, none of this exists, it doesn’t happen, it’s not part of our culture: we think we’ve been there for 5,000 years and we’ll be there forever,” she told us with a smile.
And yet, the fear is understandable and it is related to the so-called Uncanny Valley (things?): “We have an atavistic fear of what looks like us, that reminds us of us, that looks like us and reproduces our behavior. Those who have used ChatGPT or GPT-4 for a while, will have experienced some thrills, because they are software that look real”. And therefore they are a little scary, even if “the most rational people know that they are just machines, they know that the zombies in the movies are just the zombies of the movies, that it’s just a movie, even if it seems real”.
All right, then? Do we stop worrying? Nobecause the problems remain and according to Fung they are above all of two types: “The first question is that of the abuse or misuse that people will be able to make of these tools”. As also verified by othersthe problem is above all us, both because “we are the ones who decide” and because “some rules already exist, such as those that forbid using ChatGPT to make medical diagnoses or give legal opinions: if someone uses it for this, it’s not their fault of ChatGPT”.
Beyond this, and connecting to the theme of the scarce representation of even important parts of the population, the fact remains that “algorithms are imperfect” and that “whoever uses them must be informed about what they can and cannot do and whoever develops them must tell how he trained them”. According to Fung, more AIs need to be open source and “it doesn’t matter” if that of OpenAI is no longer (the company confirmed it in early March): “Google continues on this path, Hugging Face continues on this path and many others with them”. Here too, it is we (understood as us users) who can decide who to turn to for these tools and who to reward with our attention and our money.
The online project How does a genius think? Google’s AI enters the mind of Leonardo da Vinci by Vittorio Emanuele Orlando 03 July 2023
China, the EU and the USA: AI in the midst of competition
The complicated international situation increases concern, but again Fung helped us see the glass as half full: “It is true that it is a complex moment and there are divisions at the level of governments and politicians but the business world continues to work and collaborate even if maybe people don’t perceive it, and this also applies to commercial exchanges related to the field of AI”. More and more important: “The same does the scientific community, whose members continue to collaborate (somewhat like on the ISSed) and with papers they are open and accessible” and it is therefore very unlikely that China will come up with its own AI that will outwit the competition without any warning, which is a bit of a fear of the last year.
The important is that “governments are able to keep up” in this competition in which it is difficult to clearly establish who is winning: “China and the United States are undoubtedly ahead in terms of economic benefits and the so-called monetization of AI-related products, with the United States being ahead of everyone in long-term researchbecause they have a lot of money to invest”, as demonstrated by the case of Microsoft with OpenAI.
And Europe? “He does well on the longterm and there are some Countries that are working effectively, such as Francewith investments, support for startups and greater freedom of action”, but perhaps “the EU has a more conservative approach and tends to regulate more than China and the USA”.
Whether or not this will make us fall behind, it’s too early to understand. But even if it were, according to Fung we should be happy anyway: “Europe has a strong and indisputable influence from a cultural point of view, it’s not like you can excel at everything”.
@capoema