Stéphane Doncieux, Sorbonne Université

Stéphane Doncieux, Sorbonne Université

What is your opinion on Human – Centered AI?

Right now, artificial systems are mostly built to reach a certain goal and humans to some extend to adapt this machines that achieve wonderful things, but if we consider robots for an instance, for a long time they have been working alone in cages far from humans. What we are trying to work on right now is to open the cage and make new robots that could work along with humans, not replace them but work with them, to have them in their current job. There are a lot of progresses that have been done in Artificial Intelligence, thanks to different techniques. Right now, we are much more able to take into account, more complex models, a lot of issues we need to deal with, but we are making progress fast in that direction.

What would be your blue sky project in AI for Europe?

The blue sky project with unlimited resources would be first the ability to collect a lot of data, having many robots moving around interacting with humans, so that we can collect a lot of information about how it goes so that we can exploit it and then learn new representation to make robots.

Ron Chrisley, University of Sussex

Ron Chrisley, University of Sussex

What is your opinion on Human – Centered AI?

I am really excited about the Humane AI approach. Because I think the way to create the kinds of AI that we really want to see created, is to collaborate, to find collaborations between machines and people, or use machines, use AI systems to help people collaborate with each other. I think this can be seen especially in the area of creativity, which is what I’m interested in. Rather than trying to just create machines that are somehow creative, maybe it would be a better approach to build machines that can assist us in being more creative. We have some particular ideas in mind her in the Humane AI group about how to do that.

I hope that this project will gather together people who really want to see a different kind of AI in Europe, expressing the European values, human centered European values, that would be world leading. That’s why I’m most excited about, being involved in this group, because I think, not just got the right ideas, the right talent and expertise, but its heart is in the right place to. That’s what I really am looking forward to working with colleagues about.

What would be your blue sky project in AI for Europe?

What I would like to see, is, if resources weren’t an issues and time weren’t an issue, I would like to see AI systems that can not only help us and be very good at perceiving the world, but that could perceive themselves and acquire the knowledge of how they work that would allow them to be much more sophisticated that they currently are. I think the applications of such systems are many and hard to imagine would be, hard to underestimate the sophistication that one could add to AI, if this AI systems could learn about themselves with the same power that they are currently learning about photos of cats.

Wojciech Borkowski, University of Warsaw

Wojciech Borkowski, University of Warsaw

What is your opinion on Human – Centered AI?

It is important, but for me maybe more important is how to track the politics and the effects of what are people doing for the future and computer simulation is a very good tool for this. What particular political decision, how it may work in the future.

It connects?

It is very connected. In social simulation we use some kind of agents. Agents have small intelligence or less or more intelligence. We try to limit, especially because we have no tools for to make the very intelligent agents. And to make a group of very intelligent agents because of limited resources of computational resources.

Could that be for example your potential blue sky project?

First thing should be standardization of social simulation, because many people use many languages, different algorithms, it is many things work many times. The standardization will be very good and second thing will be use the language that is efficient. I use different language for simple project but different for serious project. For simple project I use processing, it is some kind of simplified Java. It works very well for students for example, but it is simple, it works on Windows, Unix, Linux and Mac. The same code work well. But for serious processes I use C++ and sometimes Quda, this is sort of parallel programming for graphic cards. If you could make something, if you could make effort to make this thing together, it would be good for this discipline I think.

Standardization of languages, programming languages?

Yes, or interaction between languages, it would be also possible to make good library in C++ and then connect the server to small module, write in Java or processing, when people that are not very good in programming, can write for example how one person, how agent work, not person, how agent work and put it into the big system.

Rinzivillo Salvatore, Research ISTI-CNR

Rinzivillo Salvatore, Research ISTI-CNR

What is your opinion on Human – Centered AI?

For the Human – Centered AI let’s say human at the center of AI is some sort of let’s say grounding this technical information that for many people might look something like “wow it’s quite magic”, but basically it is just bringing together pieces that each of us is living, some sort of digital background that we are living behind. I am doing this action of Facebook, I am buying this kind of stuff at the supermarket, I bought a flight ticket to came here in Paris for this workshop. All this kind of different pieces can be brought together in order to build our complete picture that at the moment is difficult, even for me to construct and to put together and also to make a sense of this kind of thing.

What would be your blue sky project in AI for Europe?

One challenging thing where we are starting investigating is to try to bring this kind of model where each individual in society collect some data and make sense of some data in order to bring on a common platform where each individual can make a collective intelligence, where my data may be used by the person that is near to me in order to create an intelligence, as I was saying, that can bring beneficial result for all the societies. This can be helping improving and making my own life better in terms of let’s say ecological food print, or let’s say the way that I can use my resources or my time, to improve social relationship and so on. It is a sort of meta collection of this individual data that can gives a new boost on the way we treat and sense our own data

Peter Schüller, TU Wien

Peter Schüller, TU Wien

I would say that Human – Centered AI is not a project, but is more or less prerequisite for AI to go to the future, so that AI and humans will not diverge, but will work together and continue together in a collaboration.

My blue sky project would be to have a system that can work together with humans, as I am coming from a language perspective, to interpret texts. And he big point of this system would be that it does not start from what is the correct meaning of a text, because that is different for each human, but the system would learn from different humans what it means and then it would be possible to somehow converge all to keep separate opinions about the same texts. Like the reality would require.

Bernardo Magnini, Fondazione Bruno Kessler

Bernardo Magnini, Fondazione Bruno Kessler

We are working on specific aspects Human – Centered AI, particular how the interaction between machine and the human can foster this vision of AI for human. We think that without clearly understanding of natural languages there cannot be a real interaction between machines and humans. Working in this direction is not only useful, it is necessary.

We would like to build, as I said, machines that are able to properly interact with humans. There is a long way in this direction and we are discussing in these days about next steps for this vision. Particularly, we are interested in several aspects. First of all, multilinguality. Europe is made of countries with several languages, we don’t want that there is just one language which is the reference, we want to push a research and technology for all the languages. This opens big issues, because not all languages have the same number of speakers, not each language is equally important, but we think that language is really necessary in order to preserve cultural aspects of each country and each community.

The blue sky project is a machine able to interact, to conversate on several issues without being too much forced on a specific domain on a specific language, trying to mirror as much as possible what humans do among them.

Luis Leiva, Aalto University

Luis Leiva, Aalto University

I believe as a concept, conceptually that as picking these projects is framing how users can be modeled, how the context can be modeled, how can society benefit from all the breakthrough research of this projects on the society. So far what I like the most is the framing of the projects in terms of how interaction is being considered, how the symbolic and the modern machine learning techniques are breaching together and how the ethics and transparency are being considered.

I think one of the biggest challenges is understanding the users their context. If I would have unlimited resources I would try to contribute to moral in the context. Because it is very dynamic, is very challenging and it’s a one of the long term research challenges currently going on.

Franco Turini, University in Pisa

Franco Turini, University in Pisa

What is your opinion on Human – Centered AI?

Basically, what I think is most important is the possibility of using AI in a way that is understandable for the people. Most of the risk we have now AI systems are trained using data we do not own that are collected by Google or other. That is the other problem, I mean it can be pleasant when you connect to the web, the kind of commercials you get are designed for yourself, but when for example it has to do with voting, I don’t think it is much fair for humans. Democratizing AI should be one of the goal of our projects.

What would be your blue sky project in AI for Europe?

Well essentially, I repeat myself democratizing the use of AI tools, which means, make them understandable, usable and having the basic data owned by the people itself, I mean accessible by the people itself. So it would be a project involving both computer science, but also ethics, also philosophers, also politicians, essentially, who I am not sure so far understand very well what’s going on.

Victor Paléologue, Sorbonne Université

Victor Paléologue, Sorbonne Université

What is your opinion on Human  – Centered AI?

It is really important that it is Human – centered, just because robots are only tools, it is not meant to be anything else. It is meant just to do what you want of it, you have to keep in control with your devices, would be your computer, your cellphone, your robot, it’s the same, you have to be in power with it. It has to be centered to help you personally. Its inevitable, otherwise that’s pointless to robotics. Stalking the future, we can imagine if a robot can understand whatever happens in your home, that you can tell about it, lets say about the furniture is there, about the rooms, about the people living in there and you can make everything together, lets say okay, make some meal and prepare something for the dinner for this person who will come home tonight or whatever, it is really the ideal of a robot that you can do whatever you want with it. It really needs to understand the world in which it is and just trying to make the human able to teach the robot, just this is a step towards that. In the end, the robot if it understands so much it can do whatever you like of it, like the sci-fi movies, you know. I don’t think its science fiction, its not working yet, but I don’t see why it wouldn’t happen really. Technologies are there and we are working towards that, it’s a long road, but I am pretty sure we are going to do it.

HumaneAI Futures Labs: A series of workshops for a new AI science

On June 17 and 18 the HumaneAI partners designed and host five AI Research Labs aimed at generating greater awareness of the long-term implications of  a human-centered research agenda in AI among researchers, but also aimed at crating input for policy leaders, both within and outside of the civil service. This series is supported through the 37 partners under the leadership of DFKI.

Workshops to design a Human Centered AI roadmap in Europe

HumaneAI workshops in Paris
On June 17th and 18th the project partners will meet at Centre International de Conférences Sorbonne Universités in Paris to kick-start a series of research agenda workshops to leverage EU expectations of a Humane Centered AI in Europe. Delivering a compelling research is crucial to our chances of getting long term significant impact via the new Horizon Europe programme.

HumaneAI partners contribute to OECD AI principles

HumaneAI partners working with OECD
HumaneAI partners working with OECD

HumaneAI partner Jozef Stefan Institute with expert Marko Grobelnik has been involved in designing the OECD AI Principles which are the first such principles signed up to by governments and have been discussed also in the HumaneAI project.

On 22 May 2019 the OECD member countries approved the OECD Council Recommendation on Artificial Intelligence.

Beyond OECD members, other countries including Argentina, Brazil, Colombia, Costa Rica, Peru and Romania have already adhered to the AI Principles, with further adherents welcomed. Ministerial  Meeting the AI Principles.

Relevant is the fact that this is the first AI document as a result of a political consensus among 36 OECD members and 5 non-member countries.