R & D#

Below are some projects I am involved in or that I am trying to put together.

Modulo, resources for teaching computer science#

Since 2022, computer science is mandatory in high school across Switzerland. In order to help teachers teach this course, the Canton de Vaud has funded the development of teaching resources. Together with computer science teachers, people from EPFL, UNIL and HEP, we are developing resources including a coursebook and teaching activities for teaching computer science in high school. The resources can be built as a website or as a printable handout are distributed under a Creative Commons licence. They are hosted on github, such that teachers can adapt and modify them as they see fit.

The ressources are available here.

Computer science teaching and sustainability#

What would a sustainable computer science look like and how would it be taught? Computer scientists often see themselves as providers of solutions, but may have a harder time to understand that their science may also be part of the problem, in particular when it comes to sustainablity. There is a need to examine to what extent values and ways of thinking transmitted in computer science classes are congruent with those necessary to address the challenges of our unsustainable way of life and relationship to the world. Those important questions are addressed in the phD of Baptiste de Goër, which I am co-supervising with Sophie Quinton and from the INRIA Grenoble.

Gender issues in computer science teaching#

Despite various kinds of initiatives, the relative number of females computer science students has remained quite low, at around 20% in most western societies, including Switzerland, aout half of what it used to be in the seventies and eighties. Inspired by the feminist literature on this topic, this project aims at studying gendered dynamics in the classroom and how they might affect the possibly gendered representations of technology and computer science

Teaching Latin with an AI#

This projet is spearheaded by my former student Damien Cavaleri, who trained the GPT2 text generator on a large corpus of latin texts and obtained a purposefully imperfect latin text generator . This generator can then be used in pedagogical settings, where students can analyze and correct the output of the generator. The originality of our approach lies in the fact that the AI is not an “all-knowing” or expert entity, but rather a co-learner that is making mistakes that can be analyzed. The AI can also fuel reflections about the nature of language and about the notion of correctness in a language without native speakers. Would it be possible that an AI speak better latin than any living person? Would it be advisable to entrust the AI with the power to decide what is correct and what is incorrect?

More information here.