Reasonable AI

Cluster of Excellence application for Reasonable Artifical Intelligence (RAI) submitted

The Rhine-Main Universities (RMU) have submitted their full proposals in the Cluster of Excellence (EXC) funding line of the Excellence Strategy of the German federal and state governments. TU Darmstadt has submitted full proposals for a total of three cluster initiatives, including the Reasonable Artifical Intelligence (RAI) project initiated by hessian.AI-members:

Despite groundbreaking progress in the last decade of AI research through Deep Learning (DL), current AI systems still have notable weaknesses. For one thing, they require considerable resources; for another, they cannot think logically or deal with unfamiliar situations. They do not continuously improve and need to be constantly adapted.

The new generation of AI envisioned by RAI should learn in a more “rational” way than current systems. This is done by training the models decentrally, continuously improving over time, building up an abstract knowledge of the world and having the ability to think, interact and adapt to their environment with common sense.

In addition to the RAI project, EXC full proposals were also submitted to the CoM2Life research network (led by Johannes Gutenberg University Mainz (JGU)) and The Adaptive Mind (TAM) project (joint proposal with Justus Liebig University Giessen and Phillips University Marburg).

Find out more here.

AI generated image by DALL-E.