Interview with the new board of directors member Hector Geffner


Hector Geffner joined the Board of Directors of the RWTH AI Center


As reported last year (see announcement on 28 Nov 2022), Hector Geffner joined the RWTH on 1 Jan 2023 after receiving the Alexander von Humboldt professorship.

We are now happy to announce that Hector also joined the board of directors of our AI Center. Therefore, we conducted an interview and asked him about his background, motivation, and future plans.

  Dr Julia Mann interviewing Prof Hector Geffner Copyright: © Marco Pegoraro  

First of all, congratulations for receiving the Alexander von Humboldt professorship, Hector. Could you please tell us a tiny bit about you and what your role is at the RWTH?

Thanks Julia. I'm a new professor at RWTH in the Computer Science Department, where I'll be heading the Chair of Machine Learning and Reasoning. I'm coming from Barcelona, Spain where I was a researcher at ICREA and a professor at the Universitat Pompeu Fabra (UPF). At RWTH, I'll build a team to push forward research that ties the areas of learning, representations, and planning, which are central topics in AI. I'll also teach courses on these areas, and possibly also on the social impact of technology in general. AT UPF I set up and taught for a few years a course on social and technological change that was popular and very well received by students.

That sounds exciting and I'm sure that the course on social impact of technology will be a great success. Could you please tell us more about your research interests, where can it be applied, and why you think it's relevant.

Artificial intelligence is the study of intelligent behaviour through sharp computational lenses: how can a machine do what humans do? This raises a number of very interesting questions that can be cast in a mathematical language and can be addressed computationally.

One of the key questions in AI and Machine Learning today is about the integration of learning and reasoning. In AI, we have incredible powerful deep learning systems that have revolutionized the field and have opened the door to a number of applications, but they are not reliable. This is because they rely exclusively on data; they are deep learners but shallow understanders.

In AI, we also have a number of reasoning systems or solvers that excel at reasoning and planning, but which rely on models crafted by expert hands. This dichotomy between learning from data and reasoning with models pops up in a number of contexts. Why don't we have self-driving cars in our streets? Because the systems that learn from data are not reliable, and the models needed to make sense of unanticipated situations, like a child chasing a balloon near the road, cannot be specified by hand. A key step for achieving an integration of learning and reasoning is learning from data the models that are needed for reasoning.

This state of affairs in AI has a lot to do with what Nobel Prize winner Daniel Kahneman's describes as System 1 and System 2 "thinking" in his book Thinking, fast and slow (Farrar, Straus and Giroux; 2011). In the book, System 1 refers to "intuitive intelligence", fast, reactive, and effortless, and System 2 to "analytical intelligence", slow, deliberative, and effortful. There is indeed a strong analogy between Kahneman's Systems 1 and 2, on the one hand, and learners and reasoners in AI, on the other. A crucial difference though is that our Systems 1 and 2 are tightly integrated, while AI learners and reasoners hardly talk to each other. A central challenge in AI is to get learners and reasoners to inform, enhance, and complement each other. This overall goal guides my current research which unfolds in the context of goal-directed behaviour and planning.

I think that this research will contribute to make AI systems more reliable, more transparent, and more interactive.

Thank you, Hector. What are your plans for the next few years here at RWTH?

First, I'll do my teaching and my research, contribute to the activities and visibility of the AI Center, and try to establish collaborations with RWTH colleagues. I also hope to contribute to the education of a new generation of AI scientists and engineers. For this, I will build a new team, recruiting students and postdocs. The research is funded by RWTH, the Humboldt Foundation, and an Advanced ERC Grant, and there are many funded positions to be covered. I'm excited about these research prospects because I really hope that we can make a difference on these central problems in AI and ML that are still open.

This sounds like a solid plan, and the AI Center can help you to find the right candidates.

You also joined the Board of Directors of the RWTH AI Center. What do you think will be your contribution?

I've finished my PhD at UCLA in 1989 and had the privilege of learning the craft of research from one of the best AI researchers in the history of the field: Judea Pearl, winner of the ACM Turing Award in 2011. Since then I've learned, taught, and done research in many of subareas of AI, and as a result have developed an informed, broad, and original view of the field. I hope that this view and experience, as well as my own research agenda, can contribute to make the RWTH AI Center a world-class AI Center. I'm very much aware though that my colleagues at the center, who are all world-class researchers, do not need me for this. It is indeed a privilege to be here.

You are very modest, Hector. We at the AI Center are very much looking forward to your vast experience and are confident, that your world-class knowledge will strengthen the RWTH in the field of AI and complement the expertise of other board members on our way to become one of the leading universities in AI in Germany and world-wide.

Thank you very much for your time and we are looking forward to hearing more about your research and teaching activities.