Rational Intelligence Lab: CISPA faculty publishes research group manifesto
Hardly any other topic has gained as much importance this year as machine learning. The newly founded "Rational Intelligence Lab" responds to these developments. CISPA faculty Dr. Krikamol Muandet tells us about the goal that was linked to the creation of this name: "The idea was to give ourselves a direction to work towards and having a clear vision of what are the overarching goals that we want to achieve as a research group." From there, writing a founding manifesto seemed like the next logical step. "We wanted to give people or researchers who share our vision the possibility to reach out to us and to collaborate with us and to create a community of people who strive to achieve the same goal," Krikamol continues.
According to the founding manifesto, the goal pursued by Krikamol and his three collaborators is to "address emerging mathematical and practical challenges that arise at the interface between intelligent systems and society". For Krikamol and his team, nothing less is at stake than the need “to redefine machine learning and generalisation in our modern context." The term ‘rational intelligence’, which is present in both the group's name and the manifesto, refers to the kind of systems the researchers have in mind: "In essence, our research will establish a strong foundation for creating systems that exhibit not only intelligence, but also rationality, capable of engaging in interactions with complex environments," the manifesto states.
The implications of this Krikamol explains as follows: “In certain areas, although AI applications may provide accurate predictions, they can be considered irrational by the users. Put simply, they do not make sense.“ In order to find solutions for this, it is essential to address the question of what rationality means in different contexts. To explore this, Krikamol and his team want to draw on theoretical insights from economics. According to the manifesto, the solutions envisioned by the group require "interdisciplinary fusion of computer science, statistics, machine learning, and economics".
The group's all-encompassing goal is to "further foster the accountable integration of these systems into society," as they state at the end of their manifesto. To realize this ambitious goal, they need more fellow researchers: "We always look for students, postdocs, research assistants, and collaborators who will embark on this journey together” reads the LinkedIn post in which Krikamol announced the group's formation. "People who are interested in our work can look at the manifesto and see what we are committed to," he said in conversation.
More information about the research group's work is available on its website. The manifesto can also be read in full here.