IA Generative, 5 questions (plus one) to Hiroshi Ishiguro

IA Generative, 5 questions (plus one) to Hiroshi Ishiguro

This text is part of the Special AI of Italian Techentitled "Talk to me. The artificial intelligence revolution", on newsstands free of charge with the newspaper The Republic starting Thursday 23 March

Hiroshi Ishiguro looks like straight out of a Japanese cartoon, or from a manga: black jeans, black leather jacket, bold and detached attitude. And she can definitely afford it.

On the eve of his 60th birthday, Ishiguro is an authority on robotics: lecturer at the Japanese University of Osaka, is best known for his work on androids and human-looking machines. Between 2005 and 2008 he created Repliee Q1Expo and Geminoid, respectively a robot with a female appearance and one identical to him, which occasionally teaches instead of him, to the great amazement of the students. In 2015 he developed Erica, which he defined as the "most beautiful and intelligent humanoid ever made", so much so that it has a role in the movie Sayonaraby Japanese director Koji Fukada.

The we met when he visited the Italian Institute of Technology in Genoa to sign a collaboration agreement on the development of avatars and their introduction on the market, with the aim of "improving people's lives". We met him and decided to start our chat with him in the worst possible way, asking him if artificial intelligences are female or male: he looked at us sideways, evidently perplexed. Then we explained ourselves better, specifying that in Italian the word intelligence it is feminine, which leads us to use this genre when we write or talk about it: “Now I understand - he said with a half smile - This is a problem that you Italians have, but that in Japanese and English it doesn't exist. And by the way, AI is genderless, fluid and transgender."

Future and controversy

Artificial intelligence beyond human intelligence and humans like ants: OpenAI and the risks of Strong AI

by Emanuele Capone

Since last November, everyone has been talking about ChatGPT, even those who had never been interested in AI before: are you who have been working in this field for decades, are you somehow annoyed by this?
Honestly not, also because ChatGPT isn't an artificial intelligence: it's a big database, it's like Google, nothing more than a huge statistical model. It thinks nothing, it is unable to generate anything, it is totally different from a human intelligence: it cannot elaborate or create its own concepts, but it is very good at reorganizing the concepts it has inside, on which it has studied and which it has learned.

It is still a great computing power: when will we be able to have it in the head of a robot? When will we have robots with real AI inside?
We have both too many expectations and too many fears, too many apprehensions about robots and AI, an approach that ranges from extremely positive to extremely negative: they should not be underestimated because they are able to easily solve even very complex problems and also to store, manage and reorganize large amounts of data easily. And yet, we must not get too caught up in anxiety about them. That said, robots don't need to have an AI in their heads: they are already connected to the Internet, just like we are connected through smartphones.

A big obstacle to our interaction with them is non-verbal communication: when will the robots understand our emotions?
It's true: robots are not yet able to understand our feelings from facial expressions alone, but can we humans really do it? Can we really understand each other only by the face? To understand if the other person is sad, happy, pleased or scared, just by facial expressions? It's again a problem of expectations: we expect robots to be good at something we're not good at either.

If and when they succeed, will we really be beyond the Uncanny Valley?
The Uncanny Valley concept was misunderstood: zombies are uncanny, scary, bewildering, destabilizing; a robot identical to a human and able to faithfully replicate it is not uncanny and shouldn't be scary. From this point of view, ChatGPT works very well and perhaps it is already beyond the Uncanny Valley, because it is able to sustain a conversation that resembles the human one and faithfully reproduces it. A well-made robot is this: outside it's the same but inside it's different, just like ChatGPT from the outside seems to have a mind like ours, but inside it's different.


20 Italian startups dealing with artificial intelligence

by Alessio Nisi

What world do you imagine in 2030?
In the next decade we won't need independent robots as we think of them now, able to walk and do things by themselves, but robot avatars, remotely teleoperated, inside which we can put our presence, so we can walk in distant places, be tourists , work, study, interact with students, overcome handicaps. Avatars are better than robots, because they are robots within which we are as a consciousness.


Source link