Trustworthy AI in Education

University teachers are shaping the hearts and minds of the next generation of AI researchers. It is important that PhD students in computer science and AI learn about the results of TAILOR researchers in the domain of trustworthy AI. Researchers Peter Flach and Miquel Perello-Nieto have been leading a team of researchers collecting a set of courses to cover the topic. The resulting curriculum will be free to use by any academic institution that needs guidance when developing coursework and classes for trustworthy AI.

The TAILOR curriculum promotes the teaching of trustworthy AI. It is a program that covers all the necessary topics for future professionals of AI.

– The number of AI models, methods and research has grown very rapidly in the latest decade, which makes it necessary to map AI solutions. Furthermore, AI tools will be used in critical applications that require auditability, explainability and interpretability, and need to be covered in PhD programs, says Flach, elaborating on why there is a need to coordinate these efforts.

The different courses and tools being developed have been tested at the Tailor Network summer school on three occasions. Over 800 participants have tested and provided valuable feedback which has been used to update and finetune the curriculum.

– Standardizing the educational material will facilitate the movement of students across Europe, fueling interdisciplinary collaboration of European institutions and advancing the research efforts in AI. With the rapid advancements in the field, this also needs to be a continuous cooperative process, as new courses get added to the curriculum and complementary ones get developed by different schools and universities. Ideally, it can become a toolkit and a creative commons for trustworthy AI education, continues Flach looking towards future implementations.

– Given enough adoption of the curriculum and this framework for trustworthy AI development, there is also the possibility to create specific certifications for those that have completed enough of the coursework.

But in order to build this out and implement it properly, we need more resources, explains Perello-Nieto, sharing his hope for a continuation.

In truly collaborative spirit, the work that was done within the TAILOR Network was also extended and adopted as part of the AIDA project (the International AI Doctoral Academy). Two different curricula were developed, a PhD curriculum in Trustworthy AI, which focuses on foundations of AI and advanced topics, and a second one, to show how these courses could also be used as a part of the more general curriculum in AI.

– A driving force behind this collaboration was Barry O’Sullivan, part of the steering committee of the VISION consortium which is building a network to support world-level AI education, that has developed AIDA. These efforts encompass the networks TAILOR, Elise, AI4media, and AI NET, explains Flach and Perello-Nieto, emphasizing how much further we can get when working together.

About the Researchers:
Peter Flach
Professor of Artificial Intelligence at the University of Bristol
Miquel Perello-Nieto
Research Associate at the University of Bristol